Nicolas Lopez-Galvez

Nicolas Lopez-Galvez

University of Arizona
Environmental Health Science

The objective of this dissertation project is to evaluate the renal functioning in migrant farmworkers as a function of environmental and occupational conditions in this U.S.- Mexico border region during a harvest season of table grapes. 

Objective 1: Evaluate longitudinal changes in kidney functioning over a six-month period (harvest season) in migrant farm workers of the Sonora-Arizona border region.  

Objective 2: Assessing environmental and occupational factors that influence the kidney function of migrant farm workers.

Sub-Objective I: Evaluate if episodes of heat stress accompanied by dehydration and strenuous working activities can cause repeated subclinical AKI, which over time may lead to permanent kidney injury. 

Sub-Objective II: Investigate whether exposure to pesticides in the field is affecting kidney function in migrant farm workers by dividing the 100 farm worker participants from Study Objective 1 into two groups of 50 participants (one group of participants will be only working in a certified organic, while the other group will be working in a conventional area).


Sample size (n): 150 participants (50 non-farmworkers; 100 farm workers of which 50 are organic and 50 and conventional workers).
Biomarkers of kidney via blood and urine were collected from all participants.
Heat stress measurements were collected using physical strain index (PSI), which is a validated formula based on heart rate and core body temperature. Each participant will have a data point.
Several questionnaire data. 
Study design: longitudinal (sessor data - 5 measurements in six months)
Proposed statistical analysis: Descriptive statistics will be used to summarize socio-demographic characteristics obtained from questionnaire and biological measurements. Differences between the groups will be tested using paired t-tests for continuous variables and Chi-square test for categorical variables.  To evaluate associations between variables, Pearson correlation coefficient will be used. Linear mixed effect models will be utilized to assess associations between biomarkers of kidney injury and all of the predictor variables such as occupational environmental and occupational factors. In the first set of models, farmworker (reference: non-farmworker) will be included as the primary predictor variable of interest. To complete with objective 2, another set of models will have organic farmworker (reference: non-organic worker) as the predictor variable. Also, the heat stress measurements, WBGT and PSI, will be included as a predictor variable in the last set of models. Additional predictors of interest such as smoking, hydration levels, age, blood pressure will be included in all adjusted models.   

 Summary of propose linear mixed models



Predictor Variables of Interest

Farm workers vs non-Farm workers

Model 1

Serum Creatinine*

Field work status

Adjusted Model 1

Serum Creatinine*

Model 1 adjusted for: demographics (income, education, age); behavior and health (smoking, NSAIDs, alcohol, BMI, blood pressure) 




Heat and Hydration

Model 2 

Serum Creatinine

Heat stress and hydration (PSI, water intake, soft drinks, specific gravity)

Adjusted Model 2

Serum Creatinine

Model 2 + Adjusted Model 1




Pesticides and Heat

Model 3

Serum Creatinine

Pesticide exposure (organic vs conventional)

Adjusted Model 3

Serum Creatinine

Model 3 + Adjusted Model 2


Below is the model that your class helped me last semester:


Yij = B1 + B2*I(farmeri = 1) + B3*I(farmeri = 2) + B4*timeij + B5*I*(farmeri=1)*timeij+B6*I(farmeri=2)*timeij + B7*other covariates +...+b1i + b2i*timeij + eij 

                Yij = mean kidney function (Serum Creatinine) for the ith individual at time point j

                timeij = time (in days) for the ith individual - keep as continuous/linear until able to plot data

                farmeri = categorical treatment variable.  0 = non-farmer, 1 = farmer (conventional), 2 = farmer (organic)


                other covariates = please see Table 2.

However, I am trying to look for some R code to conduct this analysis, but I am having trouble with it. Also, I was asked if I could measure the change of the outcome for every five measurement or would it be a better way to do it.
Initial Meeting
I. Who:
Client: Nicolas Lopez-Galvez (University of Arizona – School of Environmental Health Science)
Consultants: Dean, Yuankun, Wen-Wei and Sayeh (author)
II. When:
19 September 2019, 1-2pm
III. Discussion:
Our group met with Nicolas Lopez-Galvez from the school of Environmental Health Science, where he gave us more information on his research study. Nicolas had met with Stat 688 students in 2018 prior to data collection and they had proposed the model in the summary he provided online. At the time of our initial meeting, he had collected the data for his study and was in the process of cleaning the data. He thought he might be able to give the data to us to analyze for him within a week. The information he provided us is summarized below:
Migrant workers from south of Mexico are bused to Sonora, south of American/Mexican border in January to work on vineyards. The vineyards in this area are mostly organically farmed except a small portion dedicated to conventional farming. The main difference between the two methods of farming are pesticide usage.
The farmers arrived in January 2019 and stay for work in this area until August 2019, living in accommodations close to the vineyard. The farmers are mostly male and between 18 and 50 years old. Only a couple of farmers are females, therefore only males were recruited males for this study. The PSI measurements, which are ear temperature, heart rate and urine samples, are taken five times during the six months the farmers are working in the vineyard. The timing for this sampling was decided as advised by a nephrologist such that there would be a good chance to collect information on kidney function. The times for data collection as advised by the nephrologist  were the first day of arrival, third day after arrival, two weeks after arrival, 2 months after arrival and 6 months after arrival.
It was attempted to recruit 100 farmers, 50, who were involved in conventional farming and 50, from organic farming.  A control group of 50 non-farmers were recruited who lived in a nearby area 30 minutes away.  The non-farmers were office workers. The non-farmers were only tested twice (as opposed to five times for the farmers). They were tested once in February or March and once in the month of August. These months were chosen such that one would be in winter and the second test would be in the summer. 
It was attempted to test and give a questionnaire to each farmer five times in this study at the times mentioned above. All the farmers were tested the first three times but for the fourth testing period, at two months after arrival, thirty of the farmers were not present as they were working at another farm. The data is therefore lacking these individuals for the fourth test. To compensate for this issue, 20 other farmers, 10 from organic and 10 from conventional areas, were recruited and tested at two month and six month intervals. The 30 farmers that were not present for the fourth testing period, returned to the farm and it was possible to test them for the last scheduled time.
Prof. Billheimer advised that a linear mixed affects model could be suitable to use for this research. The package recommended to use in R was nlme.