.

It is usually commendable when research evidence is used to support the actions of advocacy groups. Such was the intent of the Jamaica Environment Trust (JET) recently when the results of a health survey of mining communities commissioned by them were shared with the nation.

Health surveys are commonly undertaken by epidemiologists who study the determinants and distribution of health conditions and their risk factors. I heard the results during the evening prime time discussion programmes on two radio stations. So disturbed was I by the findings and the implications for public health that I subsequently downloaded the full report, titled ‘Health Survey Analysis Mining Communities IAF Project December 21, 2015’ from JET’s website.

Sadly, the quality of the report is of such a standard that, if the reported methodology, findings and conclusions drawn go unchallenged, there may be unjustifiable and irreparable harm done to the study participants, communities, mining/quarrying companies and the nation at large.

The report was a very difficult read from the beginning, and I am left to wonder whether the corrected final draft was posted based on the many deficiencies I identified. I will briefly highlight a few of the challenges and queries I had with the main sections of the report.

Introduction

In the literature review, the researchers sought to provide evidence on the link between air pollution and a number of non-communicable diseases (NCDs) that are prevalent in Jamaica. Surprisingly, they made no reference to any of the studies previously done in mining communities in Jamaica or studies related to respiratory illnesses, such as asthma, in Jamaica. Instead, they tried to build a case for the association with NCDs, which include cardiovascular diseases (CVDs), while at the same time mentioning that “the evidence linking outdoor air pollution to cardiovascular disease in humans is still circumstantial and not smoking-gun proof — it shows correlation, not causation”.

Towards the latter part of this section the focus inexplicably shifted to obesity and a lot of statistics presented on this risk factor. In general, I found the deductive reasoning very confusing and sometimes flawed. At the end of this section I remained unclear as to what were the hypotheses, aims and/or objectives of the health survey.

Methodology & Scope

The researchers began by stating that they selected three communities located within one mile of mining/quarrying operations and two others to act as control (unexposed) communities. Apart from their locations, no other justifications were provided in the report for selecting these communities.

The authors also presented a Google image of the location of the selected exposure and non-exposure communities. However, they failed to specify in the narrative the approximate distances from each other and from the mining (exposure) facilities. Did they check historical data on the prevailing wind directions? How can we be sure that the non-exposed communities are at a distance far enough to avoid the impact by air pollution from the mining/quarrying facilities?

Inclusion criteria

The researchers mentioned the following four criteria:

1. Households from communities exposed to mining operations (downwind the prevalent wind direction, within one mile) in the areas of study

2. Households from nearby communities not exposed to mining operations.

3. Households with children under five years of age

4. Households with senior citizens

The first two criteria are mutually exclusive based on their study design. It was not clear whether it was compulsory that the households selected had to have senior citizens plus children under five years old. The age cut-off for a senior citizen was not stated. They also failed to mention the acceptable age ranges for the interviewees from each household. No exclusion criteria were mentioned.

Sample size

No information was provided to the reader on the population size of the communities and the estimated proportion of people living within them with any of the health outcomes under investigation. The latter, in particular, is a necessary input to determine if the final sample size calculation would be adequate for correctly estimating the prevalence (proportion of people) in the study population with the main health conditions being assessed.

Sampling procedure

In this section the researchers stated that a street would be considered a housing unit. I do find this difficult to conceptualise as usually a number of housing units are found on a street. They also stated that, “A random selection of dwelling units from housing units was utilized: for example, every third house on a given street.” This is an incorrect statement on sampling. What is being described is systematic and not random sampling.

Instrumentation

The survey instrument was appended in the report and included the purpose of the study: “To evaluate the effects of quarrying and mining on the communities of Hayes/New Town and Ten Miles Bull Bay in Clarendon and Kingston.”

I do not believe the study methodology was appropriate for this evaluation. ( I will further elaborate in the Results section of this critique). I also expected some mention to be made with regards to recruitment and training of the research team. For example, the researchers mentioned that they tested for hypertension and body mass index. To ensure reliability (repeatability) and validity (accuracy) of the findings, it is important to know the answers to the following questions:

* Were the correct sizes of blood pressure (BP) cuffs applied?

* Were the cuffs always placed in line with the level of the heart over a bare upper arm?

* Were individuals allowed to relax for at least five minutes before the BP was assessed?

* Did participants adopt the correct posture? An unsupported arm and crossed legs can increase blood pressure.

* Were questions asked about the use of stimulants, eg coffee, within 30 minutes of the measurements being taken?

* Was the blood pressure tests repeated if initially elevated?

* What types of scales were used to measure each person’s weight and height? Were they properly calibrated?

* What quality assurance measures were put in place for this study?

We were also not apprised as to whether the survey questionnaire was pilot-tested before final administration.

Data analysis & Ethics approval

The methodology does not state how the researchers planned to analyse the data to facilitate assessment of appropriateness. No mention was made of how potential biases that might have been introduced based on the methodology would be minimised.

For example, potential biases could have included a) selection bias based on the communities and participants selected for the study and b) information bias based on what was reported by the participants and/or the way the researchers collected the data.

There is also no indication that the research proposal was subject to ethical approval by any reputable research ethics committee or institutional review board. Among the scientific community worldwide this is the widely accepted norm. This important bit of information should have been included in the report, given the broad public health implications of this report and the fact that it is openly accessible via the Internet and there were no word limitations as is often the case when publishing in peer-reviewed journals.

I know for a fact that some of the main universities in Jamaica as well as the Ministry of Health in Jamaica abide by the principle that research of this nature must receive ethical approval prior to the commencement of data collection. Ethical approval ensures a number of things:

a) that the welfare, rights and safety of participants are protected

b) that the researcher’s right to conduct legitimate research is protected, and

c) that the reputation of institutions to which the researchers are affiliated is also protected.

Finally, a consent sheet detailing the explanatory information provided to each participant was not included in the report.

Results

The researchers listed a number of health conditions of the interviewees and family members that would be reported on. However they failed to stratify the key descriptive socio-demographic indicators by exposure and non-exposure communities to enable comparison of the differences between both groups and, importantly, to assess whether this may play a role in the measures of association (odds ratios) reported.

Specifically, we are unable to ascertain whether the communities of Hayes, New Town and the ‘control’ group Lionel Town are similar with respect to age groupings, sex, average length of residence of participants, etc. The same applies for the communities of Ten Miles, Bull Bay and Albion in St Thomas.

Seasoned researchers will readily identify that I’m making the case for consideration of the role of potential confounders, variables that may be associated with both the exposure and outcome variables but not necessary for the relationship between both along the causal pathway. This must be taken into consideration when drawing conclusions about the associations found.

The occupation of participants was omitted. How could this be done? I believe this is a major flaw in the analysis. What if individuals had been employed in the sugar industry where pre-harvest burning continues to be employed? We are aware that such methods do contribute to air pollution. The communities assessed in Clarendon are located in the sugar cane belt with the Moneymusk factory located in the Lionel Town area. The researchers should have assessed whether any of the participants from Hayes and New Town currently or previously worked in the sugar industry or other industries that increase exposure to air pollution. The same argument applies for the selected communities in the east of the island.

I’m particularly and deeply troubled by how the researchers reported on hypertension. They stated that “to simplify the comparison and calculate ORs, the comparison was made between those with SBP=>120 or DBP =>90 against those with SBP<120 and DBP <80”. Were the subsequent odds ratios calculated and reported for hypertension based on this dichotomisation? If so, then adults with prehypertension would also be included based on the criteria cited by the authors. Were blood pressure checks done on the children and adolescents who made up approximately 33 per cent of the sample? Is so, were sex-specific age and height considerations factored into the final classification of normality versus abnormality as is necessary for the paediatric and adolescent age group?

In the Executive Summary, the researchers report that they assessed the ‘incidence’ of a number of health conditions. Although cross-sectional studies may capture the incidence, which is the number of new cases of a health condition, the more appropriate term is ‘prevalence’ as both new and previously existing health conditions would be captured.

Discussion

The researchers drew conclusions about the adverse impact on the health of these communities based on findings from a study of low methodological quality. As such, I was left with too many doubts regarding the veracity of their findings. Among other things, they remarked that, “It is surprising that hypertension and prehypertension is present in 80.1 per cent of the population interviewed; 36 per cent prehypertensive and 44.1 per cent hypertensive”, yet provided no insight into the age and sex breakdown of the exposure versus non-exposure communities.

In Jamaica the prevalence of hypertension increases with age and is higher in men based on findings from the Jamaica Health and Lifestyle surveys and nationally representative surveys done in 2000 and 2008 as a collaborative venture between the Ministry of Health and the Tropical Medicine Research Unit at the University of the West Indies. They also did not calculate age or sex-adjusted odds ratios to assess if these characteristics were responsible for confounding.

Based on the aforementioned weaknesses highlighted, I would caution the study participants, communities assessed, policymakers and even JET against taking action based on the results and recommendations of this study. The report can be accessed at the following address:

www.jamentrust.org/advocacy-a-law/advocacy-law-publications/health-survey-of-mining-communities.html

Colette Cunningham-Myrie MBBS MPH MSc (Fam Med) is a lecturer in epidemiology, research methods and family medicine in the Department of Community Health and Psychiatry in the Faculty of Medical Sciences at the University of the West Indies, Mona. Send comments to the Observer or colette.cunninghammyrie@uwimona.edu.jm.