Identify the concepts, indicators, variables, and decision level for those variables

Contaminated produce was responsible for the greatest number of foodborne disease outbreaks (696) and illnesses (25,222) of any one food group in the US between 2001 and 2010 . Food handlers who practice poor hand hygiene may increase the risk of microbial transmission to produce . Our group has shown that microbial indicators on produce are associated with those on agricultural worker hands, suggesting that hand contamination contributes to the final microbial profile and load on produce .

The standard practice for reducing produce contamination risk associated with poor hand hygiene is proper hand washing (Todd, Michaels, Smith, Greig, & Bartleson, 2010), which is defined by the US FDA as scrubbing hands with soap, using potable water for rinsing, and drying hands with single-use paper or clean cloth towels. US FDA standards require that produce handlers be trained on proper hand hygiene and be provided access to appropriately hygienic facilities. However, potable, running water may be lacking in some regions. To address this challenge, an alternative waterless hand hygiene method can be the use of alcohol-based hand sanitizers (ABHS) which have been used in clinical settings since the 1970s (Centers for Disease Control and Prevention, 2002). However, ABHS studies on clinical workers cannot be generalized to ABHS effectiveness on agricultural workers' hands. Agricultural workers' hands, compared to clinical workers' hands, have different skin topography and are contaminated with soil and plant material.

Hand washing with water alone substantially reduces the prevalence of bacteria in hands (Burton etal., 2011). However, hand washing with soap is more effective in reducing the prevalence of contamination and specifically of Enterococcus spp (Burton etal., 2011). On the other hand, ABHS is significantly better than hand washing at reducing levels of Escherichia coli (Edmonds etal., 2010) and fecal streptococci (Pickering, Boehm, Mwanjali, & Davis, 2010). To our knowledge, data about the efficacy of soap and alcohol-based products to reduce Bacteroidales in hands is very limited or not available.

To date, no information about the effectiveness of hand washing with soap or SaniTwice in the agricultural production environment has been reported. The primary purpose of this study was to compare the effect of two hand hygiene interventions (hand washing with soap and a two-step ABHS intervention) on the concentrations and presence of soil and microbes on the hands of agricultural field workers. A secondary purpose was to evaluate whether any changes in soil and microbes in the two hand hygiene groups were sustained after field workers returned to harvesting produce.

On each day of data collection, approximately 40 study participants were recruited, and were randomly assigned into one of five groups: control, Two-Step ABHS, Two-Step ABHS Sustained Effect, Hand Washing, and Hand Washing Sustained Effect, as described below.

First, 3mL (two dispenser pumps) of an ABHS (disinfectant gel, active ingredient 70% ethyl alcohol, Desinfectantes y Aromatizantes, S.A., Monterrey, Mxico) was dispensed into a study participant's hands. After rubbing vigorously for 15s, participants wiped their hands with paper towels to remove excess sanitizer. An additional 1.5ml aliquot of sanitizer was dispensed and hands were rubbed again until dry.

The two groups using two-step ABHS were trained by a demonstration of appropriate technique and by completing the two-step ABHS themselves under the direction of the study staff. After this training and one cycle of harvest, the Two-Step ABHS group completed the two-step ABHS intervention immediately before their hand sampling. To assess the sustained effect of the two-step ABHS intervention, the Two-Step ABHS Sustained Effect group completed the two-step ABHS intervention before one cycle of harvest, then had their hands sampled.

First, hands were rinsed under running, potable, ambient-temperature water, which had previously been tested by the study laboratory and found to be free of coliform, Enterococcus, and E. coli bacteria. Then, 2ml (one pump of the cleanser bottle) of non-antimicrobial foam hand soap participant's hands. After rubbing vigorously for 20s, participants rinsed their hands with running, potable, ambient-temperature water, and dried them with single use paper towels. The Hand Washing and Hand Washing Sustained Effect groups were treated similarly to the corresponding Two-Step ABHS groups, except that they were trained on and performed hand washing with water and foam soap.

For all groups, hand rinsate samples were collected as follows: the study participant placed one hand in a Whirl-Pak bag containing 750ml sterile 0.1% peptone water. The worker was asked to agitate the hand in the rinsate for 30s; then a study staff member massaged the hand in the bag for an additional 30s, paying particular attention to the fingers. The study participant removed their hand, placed the second hand in the same bag, and the process was repeated. After collection, samples were placed on ice, and transported to the Laboratory of Microbial Biochemistry and Genetics at Universidad Autnoma de Nuevo Len (UANL), where they were stored at 4C until analysis. When each study participant gave a sample, study staff collected information on the participant's gender, age, and the time it took the participant to complete one cycle of harvest. Post-intervention, all study participants were asked to take an optional survey about the interventions and farm hand hygiene.

Rinsate samples (processed within 24h of collection) were inverted several times to re-suspend any particulate matter, and then an aliquot was taken for absorbance measurement at 600nm using a spectrophotometer. A second aliquot was sent overnight on ice to the Department of Food, Bioprocessing, and Nutrition Sciences, North Carolina State University for microbial source tracking analysis. Detection and quantification of the universal and human-specific 16S rDNA markers for Bacteroidales were performed using the AllBac and BFD primers and probes as described. The remaining sample was processed as follows to detect and enumerate coliforms, generic E. coli, and Enterococcus (three common, non-pathogenic bacteria), hereafter called "indicator.

Serial volumes of each sample were filtered through a 0.45m pore size cellulose filter (EMD Millipore Corporation, Billerica, MA) using a vacuum-manifold filtration system (Pall Corporation, Port Washington, NY). When filtering volumes less than 1ml, the funnel (with the vacuum closed) was pre-filled with 20ml of sterile 0.1% peptone water before the sample was added to allow even sample dispersion across the membrane prior to opening the vacuum. Following filtration through duplicate membranes for each serial volume of rinsate, each membrane was placed on a separate Petri dish containing solidified agar for bacterial enumeration. Remaining sample rinsate was stored at 4C for no more than 72h and reprocessed by filtration and plating for cases in which colony counts were inconsistent or outside of assay detection limits. The final sample rinsate volumes processed ranged from 0.01ml to 50ml.

All data, including participant demographic, survey, and sample data were entered independently by two trained individuals into separate Microsoft Excel databases compared, and reconciled by review of original laboratory forms. An additional check showed no discrepancy when 5% of the original laboratory forms were randomly selected and compared against the final database. Statistical analyses were performed using Stata 10 JMP Pro 10 and SAS 9.3 analysis using the Shapiro-Wilk test revealed that the duration of harvest, age of study participants, absorbance of hand rinsates, and log₁₀-transformed indicator bacteria concentrations were not normally distributed (data not shown). Therefore, all statistical tests used were non-parametric. In the results section, geometric mean and standard deviation are used to describe non-normally distributed data (age of study participants, duration of harvest, bacterial indicator concentrations).

Of the 159 participants, 88% were male, and the geometric mean age of study participants was 27 years (118 subjects provided age data). Gender, age, and duration of harvest cycle did not differ significantly across study groups (Table1). Fifty-one subjects agreed to answer an optional survey.

As a proxy for quantifying soil on study participants' hands, the absorbance (600nm) of hand rinsates was measured (Fig.1). The control group had hand rinsates with the highest geometric mean absorbance (A_600nm 0.236) indicating that those participants had the most soil on their hands. The Hand Washing method resulted in the cleanest hands (geometric mean absorbance A_600nm 0.005), significantly lower than all other study groups (p<0.001). The Two-Step ABHS group, compared to the control group, also had significantly less soil on hands (A_600nm 0.099, p<0.001). To evaluate the sustained effect of the two interventions, we observed that the Hand Washing Sustained Effect group (A_600nm 0.075, p<0.001), but not the Two-Step ABHS Sustained Effect group (A_600nm 0.166, p>0.05), had significantly cleaner hands than the control group.

Absorbance (600nm) of hand rinsates in study groups of farm workers harvesting produce on farms in Nuevo Len, Mexico. For each study group, the boxes display the quartiles (25th, 50th, and 75th) and whiskers extend to 1.5 times the interquartile range. Any data points outside of the whiskers are displayed individually as dots. Letters indicate significant differences by Steel-Dwass test (p<0.05).

Coliforms and Enterococcus spp. concentrations were compared in hand rinsates across all intervention groups. E. coli concentrations could not be meaningfully compared across all intervention groups because E. coli bacteria were found in only 16% of samples (25 of 159 samples, Table2).

Table2. Concentration, number and percent of samples positive for three indicator bacteria, AllBac (general fecal contamination) and BFD (human-source fecal contamination) Bacteriodales markers in hand rinsate samples from control and intervention groups harvesting produce on farms in Nuevo Len, Mexico.

Coliform bacteria were present in 100% of the control group samples and in the majority of the samples from the other groups , while Enterococcus spp. were present in all 159 samples tested. Compared to the control, the Two-Step ABHS group had a significantly lower concentration (p<0.001) and proportion of samples positive (p<0.001) for coliforms and a significantly lower concentration of Enterococcus (p<0.05). In contrast, to evaluate the sustained effect of the Two-Step ABHS, we observed that in the Two-Step ABHS Sustained Effect group, the proportion of positive samples and concentrations of these bacteria were not significantly different from the control group (p>0.05).

Interestingly, the number of hand rinsates positive for E. coli in the Hand Washing (5 of 40) and Two-Step ABHS (5 of 40) groups was similar to that of the control group (6 of 40, Table2).

As a second method for identifying fecal contamination on hands, hand rinsate samples were assayed by qPCR targeting the universal AllBac marker and the human-specific BFD marker of Bacteroidales 16S rDNA. A total of 151 hand rinse samples were processed; 8 of the 159 samples were lost during processing when sample bags broke. Sixty-six percent (99/151) of samples required dilution of DNA extracts to obtain interpretable AllBac qPCR results. Of these, one sample (Two-Step ABHS group) required dilution in excess of 10,000-fold and was reported as uninterpretable.

The universal AllBac marker was detected in 81% (123/151) of all samples. The proportion positive for the AllBac marker differed significantly by group (p<0.05, Table2). The Two-Step ABHS group (69%), compared to the control group (90%), had a significantly lower proportion of AllBac-positive samples (p<0.05). There was no significant difference in the AllBac geometric mean genome equivalent copies (GEC) per hand for any intervention group when compared to the control group (p>0.05). In contrast to the relatively high proportion of samples positive for AllBac, the human-specific BFD marker was identified in only 46% (69/151) of samples. The intervention groups, compared to the control, did not vary significantly in concentration (p>0.05) or proportion of samples positive (p>0.05) for BFD (Table2).

The primary purpose of this study was to compare the effect of two hand hygiene interventions (hand washing with soap and a two-step ABHS intervention) on the concentrations and presence of soil and microbes on the hands of agricultural field workers. Both Two-Step ABHS and Hand Washing groups, compared to the control group, had significantly lower soil on the hands. The Hand Washing group, compared to Two-Step ABHS group and the control group, had the lowest amounts of soil immediately after treatment and also had the lowest amounts of soil after participants returned to harvesting produce.

Soap acts as an emulsifier to suspend soil particles, which are removed with soap during hand rinsing, so this result is expected and consistent with previous research showing that hand washing with soap removes soil from hands. We found that the Two-Step ABHS intervention also removed soil, a result that is in opposition to most research on alcohol-based gels. The Two-Step ABHS intervention is unique in that excess sanitizer is applied and removed with paper towels during the first step. Friction created by wiping with the paper towels is the most probable reason for the observed effectiveness of the Two-step ABHS intervention at removing soil from hands.

Interestingly, only the Two-Step ABHS intervention had significantly lower indicator bacteria concentrations on hands. The Hand Washing group did not have lower proportions or concentrations of any indicator bacteria or Bacteroidales marker on hands. Previous studies have shown that hand washing removes soil, but does not effectively eliminate indicator bacteria. The foam soap used in this study is a non-germicidal soap (i.e. without antimicrobial qualities). Thus, any microbes on hands should be detectable by culture methods. In addition to soap formulation, the amount of soap, duration of hand washing, or volume of rinse water used in the intervention could have been insufficient to effect changes in amounts of indicator bacteria across groups - all of these factors have been shown to affect hand washing effectiveness. Additionally, the callused and rough surface of agricultural workers' hands and/or the high levels of soil (e.g., dirt, plant surfaces) on hands may have served to impede the removal and increase the persistence of microbes.

The Two-Step ABHS group, compared to other groups, had significantly lower proportion of samples positive for coliforms and the general Bacteroidales marker (AllBac), as well as the concentrations of both coliforms and Enterococcus spp. on hands. While bacteria were likely removed when hands were wiped with paper towels as part of the Two-step ABHS intervention, bacteria were also likely inactivated as 70% ethyl alcohol has well-documented microbicidal activity. Our results support previous research from health care settings showing that ABHS are more effective at reducing the bacterial load on hands than soaps. One efficacy study found ABHS was as effective as hand washing at removing indicator bacteria from hands after handling raw hamburger, and studies have shown increased hygiene compliance when sanitizer is available. The Two-Step ABHS group had significantly lower amounts of bacteria, compared to other groups, and this intervention may reduce the risk of disease transmission via produce resulting from harvester handling. On the other hand, neither hand hygiene method resulted in significant differences, compared to the control, on the proportion of positive samples or concentration of the human-specific Bacteroidales marker, BFD.

A secondary purpose of this study was to evaluate whether any immediate effects on microbe amounts were sustained after field workers returned to harvesting produce (20-30min; defined as harvest of 15 gallons of jalapeos transported to large collection bins located at the end of the field). Neither intervention group had significantly lower amounts of indicator bacteria, or Bacteroidales markers, compared to the control group, on the hands of farm workers after one round of harvesting produce. Previous research has shown that neither non-germicidal soap nor ethanol-based hand sanitizer leaves any residual on hands that would continue to inactivate bacteria after use. If hands are re-exposed to bacteria, as during harvesting, microorganisms may accumulate rapidly and therefore hand hygiene must be conducted frequently to prevent contamination.

The results of this field evaluation of hand hygiene interventions have several implications. First, while hand washing with non-germicidal soap resulted in significantly lower amounts of soil, the Two-Step ABHS intervention resulted in significantly lower amounts of microbiological contamination, compared to other groups. Regulatory agencies encourage the use of the most effective hand hygiene interventions and products. Thus, they might consider allowing use of alternative and effective hygiene methods such as Two-step ABHS under the right circumstances, for instance when soap and water are not readily available. Second, none of the hand hygiene methods evaluated provided sustained protection from microbial accumulation on hands after resumption of harvesting activities, meaning that hand hygiene must still be performed frequently in order to effectively control microbial cross-contamination by hands. Third, while not statistically evaluated, the presence of soil does not seem to be associated with the presence of microbes. For example, hand washing with soap, compared to Two-Step ABHS, had significantly higher soil but lower bacterial numbers.

To our knowledge, this is the first published study to evaluate hand hygiene interventions in an agricultural environment. Although sample sizes were relatively small, and the experiment was conducted on only two farms on one region of the world, appropriate statistical tests were used to ensure that the trends observed were statistically significant. One methodological limitation was our inability to sample the same hand before and after the intervention because our sampling method (i.e. rinsing a hand) biased subsequent treatments and measurements (e.g. microbes are removed from a hand). Thus, while we could not measure before and after effects of the interventions, we could compare changes in amounts of soil and bacteria across intervention groups. Further research is necessary to assess the generalizability of these results and identify promising hand hygiene approaches to effectively reduce microbial contamination on hands in the agricultural environment.