Study design, planning and preparation
MANOS is a two-country longitudinal cohort designed to collect detailed occupational exposure information over three consecutive days at the worksite at baseline (Round 1) and follow-up with participants approximately every six months for two and a half years. Each follow-up includes the collection of biological samples (urine, whole blood, and serum) and questionnaire data (demographic, behavioral, and health-related information). Research team members are located in the United States, El Salvador and Nicaragua.
MANOS protocols were approved by the Boston University Medical Campus Institutional Review Board, the Salvadoran National Ethics Committee for Health Research, and two Nicaraguan review committees within the Nicaraguan Ministry of Health: The National Ethics Committee and the Office of Teaching and Research that oversees protocol for public health investigations.
To represent populations most affected by the disease, we focused MANOS recruitment on workers from the Pacific coastal regions of El Salvador and Nicaragua. The climate on the Pacific coast is tropical, with high temperatures year-round [26]. There are two seasons; dry (November–April) and wet (May–October), between which both agricultural and non-agricultural outdoor activities often vary. We hypothesized that seasonal weather and varied work activities during the year may affect exposures and kidney function [18]. We designed MANOS to alternate data collection between seasons, with Round 1 conducted during the dry season when the agricultural harvest occurs, and employment in many non-agricultural industries at its highest.
For several years prior to MANOS, our team of investigators from Boston University, Nicaragua and El Salvador, conducted research in the region, engaging communities and becoming familiar with industry practices and worker organizations, particularly the sugarcane industry in both countries [18, 27, 28], corn cooperatives in El Salvador [29] and artisanal brickmaking in Nicaragua [11]. We also engaged leaders in the medical and public health community in the region and in the US, including researchers at the US Centers for Disease Control and Prevention [30].
To enable comparison between multiple occupational settings with differing types and levels of exposures, we sought to include workers in agricultural and non-agricultural industries in each country ranging in size from small family businesses to cooperatives and multinational corporations.
All industries and companies were selected based upon reports of CKD, willingness of the industry/company leaders to cooperate, the feasibility of following workers over time based on where workers live in relation to the worksite, and geographic location along the Pacific lowlands. Among agricultural workers, we sought to recruit participants who were manual harvesters (among sugarcane also referred to as cane cutters) and who worked with pesticides. These are the jobs thought to have the highest heat and agrichemical exposures, respectively. Within sugarcane, we sought to capture potential differences in work practices and exposures; we included sugarcane workers in three companies in Nicaragua and two in El Salvador. The companies were selected to achieve variation in company structure and work practices (e.g., amount of sugarcane harvested, number of workers employed or contracted, job specialization, work shift start/stop times and duration) and geographic area within each country.
Among brickmakers in Nicaragua, we sought oven workers in addition to workers focused on other tasks, as our prior research indicated oven work as the job with the highest CKDu risk in the industry, hypothesized to be due to extreme heat exposure [11]. Road construction workers in El Salvador did not have such clear job delineations, nor did we have hypotheses regarding high versus low heat exposure jobs, and therefore we recruited generally across the workforce.
We drafted information about the study timeline, goals, and procedures to share with industry representatives as well as expectations regarding confidentiality of worker data and our commitment not to share individual results with employers, nor to publish the names of specific employers to protect the workers against possible stigma associated with our findings. MANOS investigators in each country worked with industry leaders on logistics of recruitment and data collection to avoid interfering with work practices or participant wages.
Nicaraguan and Salvadoran investigators led the MANOS field teams –medical doctors, nurses, and bioanalysts – in carrying out the study. In January 2018, simultaneous weeklong trainings were held with Boston University investigators in both countries to review protocol, equipment, and data management procedures. Teams communicated with each other via Skype and WhatsApp, sending photos, troubleshooting equipment, and fine-tuning the protocol.
Participant recruitment
Prior to recruitment, investigators held information sessions at each worksite providing workers with an overview of the study and an opportunity to ask questions. Posters, pamphlets and PowerPoint slides with infographics were shared to describe study activities (Fig. 1).
Recruitment occurred near the beginning of the workweek to increase the likelihood of collecting three consecutive days of data for each individual. On “Day 0,” the day of recruitment, MANOS field team members arrived early at the worksite to meet with the workers as they arrived, before the start of the workday. Participants were recruited in groups of up to 20 workers/week in each country. This number was feasible for our study team to recruit and monitor each day. Workers at each worksite would gather for the initial screening and recruitment. In sugarcane in both countries, workers tended to travel from field to field in groups of 30. At each of the other worksites, we focused our recruitment activities in areas where workers would be clustered, so we could monitor them more easily over the workday.
Eligibility criteria
Workers were screened to determine their eligibility. Only male workers age 18 to 45 years were included. Age is a risk factor for CKD and we wanted to recruit men who were least likely to have CKD due to age. Participants had to have worked in their current occupation for at least one season to increase our chances of being able to follow them at the workplace in the future, and so that measured exposures could be reasonable proxies of recent exposures in prior work seasons. We excluded workers with a prior diagnosis of CKD or other related health outcomes (e.g., diabetes, HIV, hepatitis B/C, and polycystic kidney disease) to limit potential confounders of the association between exposures and decline of kidney function over time. Workers who reported hypertension (a cause and consequence of kidney disease) were excluded only if they also reported current use of medication to control hypertension and/or a recent blood pressure > 160/95 mmHg. Finally, workers with contraindications for use of the CorTemp® Sensor pill, an internal body temperature sensor, were excluded: less than 80 pounds, obstructive disease of the gastrointestinal tract, impaired gag reflex, prior gastrointestinal surgery, hypomotility of the gastrointestinal tract, and having a pacemaker or other implanted electronic medical device.
Workers deemed eligible were consented by a member of the MANOS field team who read the consent form aloud. Within the consent form were five questions that gave participants the choice to “opt-in” to: 1) receive kidney function results; 2) receive results of metals analyses indicating health risk (i.e., high concentrations of metals for which there is a reference concentration according to U.S. public health agencies); 3) have urine, blood and saliva samples saved for future research on CKD; 4) have DNA stored for future analyses; and 5) be contacted for future studies. Participants received compensation in the form of cash payments for each round of data collection.
Data collection
Round 1 questionnaire
After consenting workers, MANOS field team members administered a questionnaire on demographics, current and past occupation, work and home agrichemical use, personal protective equipment use, health and medication use, hydration practices, diet, alcohol and tobacco use, family history of CKD, and an alternative contact person. Day 0 questionnaires took approximately 45–60 minutes to administer. Days 1–3 of Round 1 data collection consisted of exposure monitoring during the work shift, pre- and post- shift physical examinations and biological sample collection, and a post-shift questionnaire (Figs. 1 and 2).
Physical examinations
Weight, blood pressure, and tympanic temperature for each participant were measured prior to and after each day’s work shift. Height was measured prior to the first day of monitoring.
Environmental heat monitoring
Waterless Wet Bulb Globe Thermometers (WBGT) measured ambient temperature, humidity, air flow, and radiant heat during the work shift. Devices were situated in work sites at locations as representative as possible of the microclimate of participants.
Personal monitoring
MANOS participants were fitted with devices to continuously monitor individual physical activity (accelerometer, right hip), heart rate (heart rate monitor, chest), and internal core body temperature (Tc) throughout the work shift. CorTemp® Ingestible Core Body Temperature Sensors are vitamin pill-sized capsule that, once ingested, move through the body’s gastrointestinal tract and wirelessly transmit Tc readings to a receiver. The receiver was fitted in the small of the back with a nylon running belt and programmed to collect Tc data in 10-second intervals.
Biological sampling
Urine samples were collected before and after the work shift on all three days. Blood samples were collected before and after the work shift on Day 3 only. Every participant also provided a saliva sample to preserve for later genetic analyses.
Post-shift questionnaire
The post-shift questionnaire asked participants about that day’s work experience (i.e., intensity, schedule, and climate), hydration, medication use, symptoms, and use of personal protective equipment (PPE), and required approximately 30 minutes to administer.
Data transfer
In each country, data from personal monitoring devices were transferred daily to secure study computers. Data from questionnaires and data collection sheets were manually entered into Research Electronic Data Capture (REDCap) hosted at Boston University (CTSI 1UL1TR001430) [31].
Biological sample processing, analysis and report-Back
MANOS investigators in each country established an indoor biological sample processing site that was cool, clean, dust-free. Upon transfer of samples from the field, lab technicians analyzed urine for osmolality with a handheld refractometer and then for specific gravity, pH, leukocytes, nitrites, protein, glucose, ketones, urobilinogen, bilirubin, and blood with urinalysis dipsticks. Optical dipstick readers were used to improve the accuracy of results. Microscopic urinalyses were also conducted.
Biological samples were stored in a − 80 °C freezers at the Agency for Agricultural Development and Health (AGDYSA) in San Salvador and at the Ministry of Health National Laboratories (CNDR) in Managua. After the completion of baseline data collection, samples were sent by international courier on dry ice to Boston University Medical Campus (BUMC) to be shipped elsewhere for analysis or retained for storage. Saliva samples were stored at ambient temperature in closed collection kits inside 50 mL protector conical tubes and sent to Beth Israel Deaconess Medical Center in Boston, MA.
Our plan was to have serum creatinine measurements from both countries at all rounds analyzed at CNDR in Nicaragua. Due to political and social instability in Nicaragua, only the Round 1 samples from Nicaragua were analyzed at CNDR. Samples from El Salvador were shipped to BUMC for analysis at Quest Diagnostics. Serum creatinine (IDMS-traceable), calcium, chloride, glucose, phosphate (as phosphorus), potassium, sodium, urea/urea nitrogen, and uric acid were analyzed at both laboratories. Total creatinine phosphokinase was additionally analyzed at CNDR and albumin and carbon dioxide were additionally analyzed at Quest Diagnostics. A subsequent validation study indicated that serum creatinine values analyzed at CNDR were comparable to those analyzed at Quest Diagnostics. Samples from subsequent rounds from both El Salvador and Nicaragua were analyzed at Quest Diagnostics.
After laboratory analyses were completed, study clinicians provided each participant an individualized report with kidney function results from pre-shift serum analyses, basic urinalysis, and hemoglobin and hematocrit values with their respective reference ranges.
Data analyses
Estimated glomerular filtration rate (eGFR) for each participant was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation for males of white or other ethnicities [32, 33]. CKD status for each participant was determined based on the two eGFRs separated by a period of at least three months recommended by the Kidney Disease Improving Global Outcomes (KDIGO) consortium [12]. Because proteinuria is not characteristic of MeN or CKDu [5], and we did not gather information about physical abnormalities of the kidney, we considered an individual to have CKDu if they presented a pre-shift eGFR < 60 mL/min/1.73m2 at both baseline (Round 1) and at a 6-month follow-up (Round 2), representing stages 3–5 CKD [12]. Participants with only Round 1 eGFR ≥60 mL/min/1.73m2 and no data in Round 2 were not considered as having CKD as per KDIGO guidelines requiring two measures of eGFR to determine disease status [12]. We assessed crude CKD prevalence by age group (18–24; 25–34; 35–45 years), family history of CKD (defined as having a father or brother with CKD), country, and industry. To minimize confounding by age, we calculated age-standardized prevalence measures using the method outlined by Rothman [34]. Age was considered categorically, and stratum-specific estimates were standardized to the age distribution across the study population. Age-standardized prevalence estimates were generated alongside 95% confidence intervals using Cochran’s formula [35].
Additionally, we used age-adjusted linear regression models to compare means between participants with and without CKD for each pre-shift serum kidney function parameter: albumin, calcium, carbon dioxide, chloride, creatinine, creatine phosphokinase, glucose, urea nitrogen, phosphorous, potassium, sodium and uric acid. Age was treated as a categorical measure. All analyses were conducted in Stata IC 16.1. The dstdize package was used to calculate age-standardized prevalence [36].