Microlevel activity time series (MLATS) data were gathered on hand contact activities of 38 children (1–6 years old) by videotaping in primarily outdoor residential environments. The videotape recordings were then translated into text files using a specialized software called VirtualTimingDevice™. Contact frequency (contacts/h), duration per contact (s/contact), and hourly contact duration (min/h) were summarized for outdoor hand contacts with 15 distinct object/surface categories (“Animal”, “Body”, “Clothes/Towels”, “Fabric”, “Floor”, “Food”, “Footwear”, “Metal”, “Non-dietary Water”, “Paper/Wrapper”, “Plastic”, “Rock/Brick”, “Toys”, “Vegetation/Grass”, and “Wood”) and two aggregate object/surface categories (“Non-dietary objects/surfaces” and “Total objects/surfaces”). For outdoor both hand contacts with “Total objects/surfaces”, contact frequencies ranged from 229.9 to 1517.7 contacts/h, median durations/contact ranged from <1 to 5 s, and hourly contact durations ranged from 42.6 to 102.2 m/h.
The data were analyzed for significant differences in hand contact activities as a function of (1) age, (2) location, (3) gender, and (4) hand. Significant differences (P⩽0.05) were found for all four factors analyzed. Hourly contact durations with “Non-dietary objects/surfaces” and “Total objects/surfaces” increased with age (P=0.01, rs=0.42 and P=0.005, rs=0.46, respectively), while contact frequencies and hourly contact durations with “Wood” decreased with age (P=0.02, rs=−0.38 and P=0.05, rs=−0.32, respectively). Location was found to affect contact frequencies and hourly contact durations with certain objects/surfaces. For example, contact frequencies and hourly contact durations with “Fabric” were higher indoors (P=0.02 for both), while contact frequencies and hourly contact durations with “Vegetation/Grass” were higher outdoors (P=0.02 and P=0.04, respectively). Girls had longer hourly contact durations with “Footwear” (P=0.02), “Non-dietary objects/surfaces” (P=0.03), and “Total objects/surfaces” (P=0.01) than boys. The right hand had longer hourly contact durations with objects that are often manipulated with the hand (e.g., “Toys” (P=0.0002)), while the left hand had longer hourly contact durations with passively touched objects/surfaces (e.g., “Clothes/Towels” (P=0.003) and “Floor” (P=0.04)).
Information about hand contact activities can help researchers understand how contaminants load onto the surface of the hands. These chemicals may then enter the body via dermal absorption or non-dietary ingestion. Microlevel activity time series (MLATS) data capture the sequence and duration of contact events and allow contact frequency to be calculated (Ferguson et al., 2005). MLATS data can be combined with environmental concentrations to create sequential time exposure profiles that help researchers determine contact activities that lead to peak exposure (Ferguson, 2003; Riley et al., 2004).
Microlevel activity data are currently used in physical–stochastic models which can account for temporal variability in human exposure (Zartarian and Leckie, 1998). Models include the Stochastic Human Exposure and Dose Simulation (SHEDS) model developed at the United States Environmental Protection Agency (USEPA) (Zartarian et al., 2000), as well as the Dermal Exposure Reduction Model (DERM) (Zartarian, 1996) and the Cumulative and Aggregate Simulation of Exposure (CASE) model developed at Stanford University (Canales, 2004).
Despite their usefulness, microlevel activity data are time- and labor-intensive to collect. Thus, only a small number of published studies on the collection and analysis of hand microlevel activity data are available (Zartarian et al., 1997a; Reed et al., 1999; Freeman et al., 2001, 2005). Furthermore, previous studies collected data in primarily indoor environments. This study aims to contribute to the field of exposure assessment by providing an analysis of hand MLATS data collected in primarily outdoor environments. This information may be useful in assessing children's exposure to contaminants in outdoor environments. Statistical tests were applied to analyze whether hand contact activities differed by age, gender, indoor/outdoor environments, and left/right hands. Similar analyses of hand-to-mouth and other mouthing contacts can be found in AuYeung et al. (2004).
Recruitment of Subjects
Subjects were recruited from the southern region of the San Francisco Bay peninsula by calling telephone numbers randomly extracted from the Pacific Bell residential telephone directory for a 300–400 square mile region associated with Redwood City, Menlo Park, and Palo Alto. To screen for middle-class children, a phone survey was administered, and families with children between 1 and 6 years of age who lived in a residence with a lawn and whose annual household income was greater than $35,000 were invited to participate in the studies.
Stanford University's Exposure Research Group videotaped the 38 children between August 1998 and May 1999. The subjects were videotaped only on days with fair weather (i.e., no rain). Parents were asked to keep their child outdoors for the duration of the videotaping session (2 h) and videotape data were collected for each child during “natural play” (i.e., unstructured play). In most cases, only one child was videotaped during each session. In rare cases where two children from the same household participated in the study, two teams of researchers collected videotaping data in the same session. Outdoor locations where videotaping occurred include parks, playgrounds, and outdoor areas of the children's homes (e.g., back yard, front yard, patio).
Video-Translation Using VirtualTimingDevice™
A hand contact was defined as any contact with the fingers, palm, and/or back of the hands. Using a specialized software package developed at Stanford University (VirtualTimingDevice™), the videotapes were “translated” or converted into plain text files with sequential, second-by-second information on microenvironments visited and objects/surfaces contacted (Zartarian et al., 1997b; Ferguson et al., 2005). The videotapes were translated separately for the left and right hands. Contacts lasting less than 1 s were recorded as zeros. Since contacts of zero duration represent short but real contacts, they were included in the data analysis. The location grid in the VirtualTimingDevice™ palette (Figure 1) used in this study was designed to capture the different outdoor locations where children were likely to visit. The 36 object categories in the objects/surfaces grid were developed based on assumptions of objects/surfaces that children most commonly come into contact with in outdoor environments. The contact type grid allowed researchers to designate contacts as either constant (e.g., holding a ball) or repetitive (e.g., bouncing a ball). The “Repetitive” cell was selected when contacts with an object/surface occurred at a high rate.
Quality Control of Video-Translation
Rigorous training and quality control protocols were applied to ensure high quality data (Ferguson et al., 2005). Between 5 and 10% of the total number of translated files were randomly selected to be checked by an experienced researcher for agreement in object and location designations. In general, if the two files had greater than 10% difference in total duration with different object/location combinations, then the translator had to retranslate the videotape segment. If systemic errors were made (e.g., errors in the categorizations of locations and objects/surfaces), then all segments where the errors occurred were retranslated. The quality control process reiterated until the retranslated segments met the quality control standard.
In addition, as part of a separate study, researchers rewatched the videotapes to assign contact configurations for all hand contacts. Each recorded hand contact was examined and a contact configuration (e.g., open hand grip, palm only contact, etc.) was assigned to the contact. During this process, any additional errors in location and object/surface designations found in the activity files were corrected.
The translated plain text files were processed before the data were analyzed. First, repetitive contacts were assumed to alternate between contacts with the object/surface and air (i.e., “Nothing”). The contacts were assumed to alternate at a rate of 0.5 s with “Nothing” and 0.5 s with the object/surface. Secondly, the durations of continuous contacts in different locations were joined to avoid overcounting the number of contacts (Figure 2).
Object and location designations from the VirtualTimingDevice™ palette (Figure 1) were grouped to form two location and 15 object/surface super-categories (Table 1). The “Other” location and “Other” object/surface categories were not analyzed since they comprised less than 0.03% and 0.2%, respectively, of the data set's total duration. In grouping the object/surface categories into super-categories, the surface type (e.g., “Metal”, “Plastic”, and “Wood”) was assumed to be a more important factor in differentiating the concentrations of contaminants on an object/surface than the object type (i.e., “Tool/Appliance” vs. “Wall/Furniture”). Therefore, the objects were grouped by surface type. The only exception to this rule was objects in the “Toys” super-category. All toys regardless of their surface types were grouped into one super-category (i.e., “Toys”) because they are objects of contact unique to children and chemical residue data are typically available for this category of objects.
Contact frequencies, median duration per contact, and hourly contact durations were calculated for each child, hand (i.e., left and right hands), object/surface super-category, and location super-category (i.e., “Indoor” and “Outdoor”) combination. Time in view was calculated as the duration of a videotape translation subtracted by the total duration designated as “Not In View”. Contact frequencies and hourly contact durations for the left and right hands were calculated separately following Equations 1 and 2, respectively. Contact frequencies and hourly contact durations for combined left and right hands (i.e., both hands) were calculated by adding together the contact frequencies or hourly contact durations for the left and right hands.
Equation 1. Contact frequency for left and right hands
Equation 2. Hourly contact duration for left and right hands
Contact frequency, median duration per contact, and hourly contact duration were analyzed for significant differences between age, gender, right/left hands, and indoor/outdoor locations. Tree analysis was applied in S-PLUS to test whether the data should be grouped by age and non-parametric tests were used for all statistical analyses. The two-tailed Wilcoxon rank sum test was applied to analyze differences between male and female subjects, while the two-tailed Wilcoxon signed rank test for matched pairs was applied to test for differences between the right/left hands and between indoor/outdoor environments (Rice, 1995). Spearman's rank correlation coefficient was calculated to test for correlations between hand contact activities and age (in months). Only outdoor data were included in the statistical analysis of differences in hand contact activities as a function of age, gender, and left/right hands on hand contact activities. Except when the data were analyzed for differences between the left and right hands, all statistical tests were applied only on data for both hands combined.
In all, 38 children (1–6 years old) were recruited (Table 2). A total of 30 children were recruited based on satisfaction of all selection criteria and due to time constraints, an additional eight children were recruited through acquaintances. The only selection criteria for these eight children were that the children were between 1 and 6 years of age, lived in Redwood City, Menlo Park, or Palo Alto, and the families were willing to participate in the studies.
All of the subjects lived in a suburban setting. The median annual household income was greater than $100,000. Of the 31 households that provided ethnicity data, 55% were Caucasian, 16% were Mexican American/Latino, 6% were African American, and 23% belonged to other ethnicities.
Unlike hand-to-mouth contact behavior which have been found to “naturally” separate into ⩽24 and >24 months old age bins (Tulve et al., 2002; AuYeung et al., 2004), no age groupings were observed for hand contact activities. Instead, the data were grouped by “Indoor” and “Outdoor” locations because recent studies (Freeman et al., 2001; AuYeung et al., 2004; Black et al., 2005) found differences in hand-to-mouth contact frequencies between indoor and outdoor environments.
Summary statistics of hand-to-object/surface contact frequency, median duration per contact, and hourly contact duration for outdoor environments are provided in Tables 3, 4 and 5. For each object/surface super-category, contact frequency (Table 3) and hourly contact duration (Table 5) were calculated for all children (n=38), while median duration per contact (Table 4) could only be calculated for those children who had contacts with that object/surface.
For the data analyses, the median time in view was 1 h and 40 min for outdoor environments and 23 min for indoor environments. As this study focused on collecting outdoor hand contact activities, only nine children in the data set spent 15 or more minutes in view indoors. All indoor data analysis were performed using only data from these nine children, six of whom were between 4 and 6 years of age, and three of whom were between 1 and 2 years of age.
Contact frequencies, median contact durations, and hourly contact durations were analyzed for significant differences as a function of (1) age, (2) location, (3) gender, and (4) hand. No significant differences or correlations (P⩽0.05) in median contact durations were found for any of the four factors analyzed. However, significant findings were observed for contact frequencies and hourly contact durations for all four factors analyzed.
Age (in months) was found to have significant correlations with hand contact activity. Hourly contact duration increased with age for contacts with “Non-dietary objects/surfaces” (P=0.01, rs=0.42) and “Total objects/surfaces” (P=0.005, rs=0.46), but decreased with age for contacts with “Wood” (P=0.05, rs=−0.32). Contact frequency also decreased with age for contacts with “Wood” (P=0.02, rs=−0.38).
Differences between indoor/outdoor environments were examined using only data from the nine children who spent more than 15 min indoors. Differences between locations in contact frequencies for both hands combined were found for contacts with “Body”, “Fabric”, “Food”, “Paper/Wrapper”, “Vegetation/Grass”, and “Wood”. Differences between locations in hourly contact duration for both hands combined were observed for contacts with “Body”, “Fabric”, “Food”, “Vegetation/Grass”, and “Wood” (Table 6).
Differences between the genders were found for both hand outdoor contact frequencies with “Body” and both hand outdoor hourly contact durations with “Footwear”, “Non-dietary objects/surfaces”, and “Total objects/surfaces” (Table 7).
Differences between the left and right hands were found for outdoor contact frequencies with “Clothes/Towels” and outdoor hourly contact durations with “Clothes/Towels”, “Floor”, “Toys”, “Vegetation/Grass”, “Non-dietary objects/surfaces”, and “Total objects/surfaces” (Table 8).
The videotaping and video-translation methods were used to collect microlevel activity data because the duration of most hand contacts is very short (on the order of a few seconds) (Zartarian et al., 1997a; Freeman et al., 2005), and events of such short duration may not be accurately captured via other direct observational techniques like parental observation or record-keeping (Timmer et al., 1985). Furthermore, the videotaping and video-translation methods allow researchers to retain a permanent record of the raw data, which can be used for further analysis should future needs arise. It also allows researchers to retranslate videotape segments and compare the translated files in order to ensure high quality data.
For both hand contacts with “Total objects/surfaces”, contact frequencies and hourly contact durations were very high (229.9–1517.7 contacts per hour and 42.6–102.2 min/h, respectively), while the median durations per contact were very short (<1 to 5 s/contact). Thus, children have a large number of short contacts. The highest median contact frequencies were with “Toys” (129.4 contacts/h for both hands combined) followed by “Clothes/Towels” (65.7 contacts/h for both hands combined) and “Body” (65.1 contacts/h for both hands combined). This is not surprising since nearly all of the children played with some type of toy during the time they were videotaped and certain play activities involving toys (e.g., bouncing a ball or tossing a frisbee) involved high contact frequencies with the hands. Contact frequencies with “Body” and “Clothes/Towels” were also high since all children had opportunities to contact these objects/surfaces at all times.
As this study collected hand contact data mainly from outdoor environments, some of the significant findings regarding indoor/outdoor differences may be biased due to the limited amount of indoor data collected. For example, in this study, the children spent a large proportion of their observed indoor time engaging in eating/drinking events as a break from their outdoor activities. Therefore, the significant differences in hand-to-food contact behaviors between indoor/outdoor locations may be an artefact of this bias towards recording an unusually high frequency of indoor eating events. Despite the limitation just mentioned, other significant differences between indoors/outdoors environments seemed to make intuitive sense. That is, contact frequencies and hourly contact durations with “Fabric” were higher indoors than outdoors because curtains and sofas were more likely to be found indoors than outdoors, while contact frequencies and hourly contact durations with “Vegetation/Grass” were higher outdoors because plants were more likely to be found outdoors than indoors.
The right hand was found to have longer hourly contact durations with objects that are often manipulated with the hand (i.e., “Toys”). On the other hand, the left hand had longer hourly contact durations with objects/surfaces that did not need to be manipulated (e.g., “Clothes/Towels” and “Floor”). An explanation for this is that even though the left hand was not used as often as the right hand to manipulate objects, this did not mean that the left hand did not touch objects/surfaces. The left hands of the children were often observed to rest on or touch objects/surfaces such as a table, clothes, the floor, etc. Occasionally, the left hand was involved with object manipulation as well (e.g., carrying a big ball). It is important to study the two hands separately because differences in contact behaviors between the left and right hands may lead to differences in contaminant loadings between the hands. This in turn may affect estimations of non-dietary ingestion exposure from hand-to-mouth contacts if young children preferentially mouth one of their hands.
Comparisons with Previous Studies
Table 9 compares results from this study with results from previous studies. As all previous studies focused on indoor environments, only the indoor data from the nine children who spent more than 15 min in view indoors are presented in the table. It is worth noting that due to a number of factors such as differences in data collection methods, object/surface categorizations, as well as the subpopulations studied (both in terms of the ages of the children studied and the regions/locations where these studies were conducted), it is difficult to compare results from different studies.
Zartarian et al. (1997a) collected hand-to-surface contact frequency and duration data in the Salinas Valley of California by videotaping four children in a primarily indoor home environment and then transcribing the videotapes with the VirtualTimingDevice™ software (previously known as VideoTraq). Reed et al. (1999) videotaped a total of 30 children aged 2–6 years in either a day-care center or a residence in New Brunswick, New Jersey. Hand-to-surface contact frequencies were then collected by manually recording the frequency over 5-min intervals. Freeman et al. (2001) employed the same data collection method as Reed and collected data from 19 children in the residential home environment. Freeman et al. (2005) videotaped 10 children for approximately 4 h each in a primarily indoor home environment and translated the videotapes using the VirtualTimingDevice™ software.
Despite the fact that only nine children were included for indoor data analysis, indoor results from this study seemed to agree well with those from previous studies. Contact frequencies calculated from this study were systematically higher than those reported by Reed et al. (1999) and Freeman et al. (2001), but systematically lower than those reported by Freeman et al. (2005) (except for “Carpet” and “Textured Surface”). A possible explanation for the lower contact frequency reported by Freeman et al. (2005) for “Textured Surface” is that their “Textured Surface” category did not include contacts with “Carpet” or “Upholstered Furniture”, while contacts with these objects/surfaces were included as “Textured Surface” for this study as well as the studies by Reed et al. (1999) and Freeman et al. (2001). Flooring type may have contributed in part to differences in contact frequencies with “Hard Floor” and “Carpet” between this study and the study conducted by Zartarian et al. (1997a). That is, contact frequency with “Hard Floor” from this study is similar to contact frequency with “Carpet” reported by Zartarian et al. (1997a) and vice versa.
Since the children in this study were videotaped for only 2 h each and they were told by their parents to play outdoors for the duration of the videotaping session, their activities may not be entirely representative of how children spend time outdoors. However, the data collected from this study provides a rough description of children's hand-to-object/surface contacts in outdoor residential environments. In a future study, the videotapes of the 38 children will be rewatched and descriptions of general play activities will be added to the MLATS file. Then, the MLATS data will reanalyzed on an activity-specific basis (e.g., playing on playground structure, riding bicycle, gardening) to further refine information on children's hand-to-object/surface contacts in outdoor residential environments.
The results of this study suggest that in outdoor residential settings, young children have a large number (median frequency is >500 contacts/h for both hands combined) of short contacts (<5 s) with objects/surfaces. Differences in hand contact activities were found for all four factors studied: age (in months), gender, left/right hands, and indoor/outdoor environments. Besides providing an account of factors that affect hand contact activities, this study may also help improve outdoor exposure assessments by providing detailed summaries of hand-to-object/surface microlevel activity data collected from outdoor environments.
AuYeung W., Canales R.A., Beamer P., Ferguson A., and Leckie J.O. Young children's mouthing behavior: an observational study via videotaping in a primarily outdoor residential setting. J Children's Health 2004: 2 (3–4): 271–295.
Black K., Shalat S.L., Freeman N.C.G., Jimenez M., Donnelly K.C., and Calvin J.A. Children's mouthing and food-handling behavior in an agricultural community on the US/Mexico border. J Expos Anal Environ Epidemiol 2005: 15 (3): 244–251.
Canales R.A. The cumulative and aggregate simulation of exposure framework. Ph.D. Thesis Dissertation, Stanford University, 2004.
Ferguson A.C. Estimating percutaneous absorption using micro-level activity time series (MLATS). Ph.D. Thesis Dissertation, Stanford University, 2003.
Ferguson A.C., Canales R.A., Paloma B., AuYeung W., Key M., Munninghoff A., Lee K.T., Robertson A., and Leckie J.O. Collecting children's activity patterns using videotaping and video-translation. J Expos Anal Environ Epidemiol 2005 (advance online publication) October 12, 2005.
Freeman N.C.G., Hore P., Black K., Jimenez M., Sheldon L., Tulve N., and Lioy P.J. Contributions of children's activities to pesticide hand loadings following residential pesticide application. J Expos Anal Environ Epidemiol 2005: 15 (1): 81–88.
Freeman N.C.G., Jimenez M., Reed K.J., Gurunathan S., Edwards R.D., and Lioy P.J. Quantitative analysis of children's micro activity patterns: The Minnesota Children's Pesticide Exposure Study. J Expos Anal Environ Epidemiol 2001: 11 (6): 501–509.
Reed K.J., Jimenez M., Freeman N.C., and Lioy P.J. Quantification of children's hand and mouthing activities through a videotaping methodology. J Expos Anal Environ Epidemiol 1999: 9 (5): 513–520.
Rice J. Mathematical Statistics and Data Analysis, 2nd edn. Wadsworth, Belmont, CA, 1995.
Riley W.J., McKone T.E., and Cohen Hubal EA Estimating contaminant dose for intermittent dermal contact: Model development, testing, and application. Risk Anal 2004: 24 (1): 73–85.
Timmer S.G., Eccles J., and O'Brien K. How children use time. In: Juster F.T., and Stafford F.P. (Eds.). Times, Goods, and Well-Being. Survey Research Center, Institute for Social Research. Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor, MI, 1985.
Tulve N.D., Suggs J.C., McCurdy T., Cohen Hubal E.A., and Moya J. Frequency of mouthing behavior in young children. J Expos Anal Environ Epidemiol 2002: 12 (4): 259–264.
Zartarian V.G., Ferguson A.C., and Leckie J.O. Quantified dermal activity data from a four-child pilot field study. J Expos Anal Environ Epidemiol 1997a: 7 (4): 543–552.
Zartarian V.G., Ferguson A.C., Ong C.G., and Leckie J.O. Quantifying videotaped activity patterns: Video translation software and training methodologies. J Expos Anal Environ Epidemiol 1997b: 7 (4): 535–542.
Zartarian V.G., and Leckie J.O. Dermal exposure: the missing link. Environ Sci Technol 1998: 32 (5): 134A–137A.
Zartarian V.G., Ozkaynak H., Burke J.M., Zufall M.J., Rigas M.L., and Furtaw E.J. A modeling framework for estimating children's residential exposure and dose to chlorpyrifos via dermal residue contact and non-dietary ingestion. Environ Health Perspect 2000: 108 (6): 505–514.
Zartarian V.G. A physical-stochastic model for understanding dermal exposure to chemicals. Ph.D. Thesis Dissertation, Stanford University, 1996.
The authors specially thank Kelly Naylor, Nolan Cabrera, Sandy Robertson, Kevin Lee, Amy Munninghoff, Veronica Vieira, Jessica Ramirez, and Angela Lin for their invaluable help in collecting the data presented in this paper. This project was supported by EPA STAR grant #R82936201, ORETF Study #ORF018, USEPA Contract #QT-RT-99-001182, and UPS Foundation grant #2DDA103. This research has not been subject to federal peer and policy review and therefore does not necessarily reflect the views of the funding agencies. No official endorsement should be inferred.
About this article
Cite this article
AuYeung, W., Canales, R., Beamer, P. et al. Young children's hand contact activities: An observational study via videotaping in primarily outdoor residential settings. J Expo Sci Environ Epidemiol 16, 434–446 (2006). https://doi.org/10.1038/sj.jes.7500480
- dermal exposure
- non-dietary exposure
- activity pattern
- hand contact behavior
- microlevel activity
Mouthing activity data for children age 3 to <6 years old and fraction of hand area mouthed for children age <6 years old in Taiwan
Journal of Exposure Science & Environmental Epidemiology (2018)
Journal of Exposure Science & Environmental Epidemiology (2015)
Journal of Exposure Science & Environmental Epidemiology (2012)
Journal of Exposure Science & Environmental Epidemiology (2009)