How to determine whether a population has unique characteristics that require measures adaptation?
It can be challenging to know for certain a priori whether a target population has unique characteristics that require measure adaptation. This case study presents an intervention in which researchers are interested in understanding if an intervention that addresses barriers to drinking tap water at home will have an impact on beverage intake of parents and children (6 months to 3 years old).
The study seeks to understand and address the barriers to drinking tap water instead of sugary drinks in a community in the United States that is 80% Hispanic (predominantly from El Salvador and Guatemala), foreign-born (68%), and low-income (1 in 4 children live in poverty), with 15% and 26% of adults self-reporting being diabetic and obese/overweight, respectively. Parents reported that their infants and toddlers (<3 years) drank 79 g of sugary drinks per day.
To develop the intervention, the researchers engaged with the community to document how living in an environment saturated with sugary drink options and limited access to clean, palatable tap water can contribute to increased sugar-sweetened beverages (SSB) consumption, and how sociocultural norms reinforce SSB consumption in a Central American immigrant community.
Before determining a measure, the research team carried out formative qualitative work, which described how residents in this community did not drink tap water because they did not like the taste or perceived that it was unsafe. This work described how mothers were knowledgeable of the health benefits of drinking water, but food customs and heavy promotion enticed them to consume SSBs. A community survey documented that before migrating to the United States, 52% of respondents drank tap water because it was convenient, they liked the taste, and perceived that it was safe and economical. Once in the United States, only 9% of these same respondents reported that they drank tap water because of safety concerns, taste, and lack of convenience. To determine whether the intervention is successful, the researchers need a measure of beverage intake.
Challenges and Considerations
What is an appropriate dietary instrument to test sugary drink consumption for low-income, low-literacy parents and their young (6 month–3 year old) children?
The dietary instrument needs to be administered quickly but also be sensitive enough to pick up any preliminary findings for intervention impact. Additionally, the instrument needs to be understood by the colloquial Spanish of the population. While multiple 24-hour dietary recalls are the gold standard to assess dietary intake, the data collected would likely yield too much information, especially on foods consumed when the research team is only interested in beverages. Recalls are also a burden on participants’ time and require considerable resources for analysis.
Does a brief measure exist that can be used in both infants and toddlers?
The research team can use NCCOR’s Measures Registry to locate screeners. One option is a screener that has been previously developed only for beverages and validated for Hispanic preschoolers (Lora 2016). However, in reviewing this screener, the research team does not know if or how the types or quantities of beverages would be different for the study’s age group and Central American Hispanic population.
Does the measure capture all sources of sugar consumed by this population?
Anecdotally and from working with the community advisory board, researchers suspect that a main source of sugary may be traditional sweeteners (honey or panela—raw sugar blocks) added to plain water. The study population does not typically consider this a sugary drink, so it may not be captured using the traditional categories of a screener.
From the formative work, the research team determined that the population appeared to have unique characteristics that would require adaptation of the brief screener found in the Measures Registry, which was developed and validated for Hispanic preschoolers
To address the challenge of different quantities, accuracy of added sugars in recipes, and assessment of other beverages that may not be captured by the brief instrument, the research team conducted repeated 24-hour dietary recalls (using interviewer-administered ASA-24) on a subsample to compare to the responses from the beverage screener.
To assist with the low-literacy and participant burden, the beverage screener was interviewer-administered using aids that included pictures of typical drinks and their quantities, and models of different sippy cups, bottles, and cups to better assess quantities.
To determine if the population understood the survey items, instruments and methods of administering them were cognitively tested with a convenience sample from the same population. This was done to adapt language to more colloquial Spanish.
To address the concern about additional beverages that may contribute to sugary drink intake, local examples obtained from the 24-hour recalls, consultation with mothers, and the community advisory board were added to the brief beverage screener. For example, in fruit-flavored drinks, the following examples were added: homemade lemonade and aguas, horchatas, tamarind juices, and mango nectar. Examples of familiar brands and types of beverages within each beverage category were also added. For example, participants were unable to distinguish 100% fruit juice from a fruit-flavored drink marketed as made with real fruit or with vitamin C, but they would be able to distinguish Tampico (an orange-flavored drink) from a brand of 100% orange fruit juice. In addition, the team added an “Other” category to capture information on other drinks that the data collector could not fit into established categories of the beverage screener for preschoolers. These drinks included toddler milks and prepackaged yogurt beverages. The brands and quantities consumed of these prepackaged toddler milks and yogurt beverages were collected via pictures during the beverage screening to assess later during analysis if the instrument must be adapted and how.
- When studying new target populations or communities, all instruments should be cognitively tested to assess if, the question elicits the same concept that the researchers intended to understand in the target population.
- Even after cognitively testing an instrument, it is challenging to know for certain a priori if the target population will be meaningfully different from the reference population for which the instrument was prepared. At best, there is formative work and anecdotal evidence that suggest that the populations may be significantly different.
- Sometimes it is not until the data are being analyzed that researchers may notice that the original instrument does not capture important data for the new study population. For example, in the current case study, if the research team had administered the short instrument without any adaptations and then compared results against multiple 24-hour recalls (gold standard), they would have noticed that the original shorter instrument missed important information about sources of sugary beverages that the study population consumed. Alternatively, if the team had triangulated with observations in the home or had shared results with a community advisory board or knowledgeable key informants, they might have learned that they were missing important sources of sugary beverages in the study population. Similarly, if the instrument is a psychometric scale, and the study population is meaningfully different from the reference population, then the Cronbach α for the scales may not load in the same way as for the reference population. These are all examples that suggest that the target population is meaningfully different from the reference population and that the instruments will therefore require careful adaptations (beyond cognitive interviews) to improve their “fit” to the new target populations’ context.
- Colon-Ramos U, et al., How Latina mothers navigate a ‘food swamp’ to feed their children: a photovoice approach. Public Health Nutr, 2017;20(11):1941-1952. doi:10.1017/S1368980017000738
- Fuster M and Colon-Ramos U, Changing Places, Changing Plates? A Binational Comparison of Barriers and Facilitators to Healthful Eating Among Central American Communities. J Immigr Minor Health. [published correction appears in J Immigr Minor Health. 2017 May 11;:]. J Immigr Minor Health. 2018;20(3):705-710. doi:10.1007/s10903-017-0588-2