Role of ethnicity and residency in active tuberculosis in Karakalpakstan: study protocol of a matched case-control study

Abstract

Introduction

Uzbekistan is one of the 30 countries with the highest tuberculosis (TB) morbidity, accounting for 87% of all cases globally. The incidence of TB in the Republic of Karakalpakstan (RK) remains consistently high in comparison with other regions of Uzbekistan. The incidence rates of TB strongly differ even among the regions within RK. In 2019, the highest rates were registered in the northern and north-western regions, and the lowest in the southern regions. An important issue is the extent to which specific residencies impact individual health. The ethnic composition of the population of RK can be one of the possible causes for such geographical heterogeneity in TB morbidity.

Methods and analysis

The case-control design of this study primarily aims to evaluate the role of ethnicity and residency on the development of active TB in the community living in RK. Additionally, the study aims to measure the association between factors of TB morbidity and ethnicity/residency in the community of RK. It was decided to include one matched control for every included case, to control potential confounders. Cases with clinically diagnosed or bacteriologically confirmed TB will be randomly sampled from the patient registry. Controls will comprise individuals without TB, living in RK at the same time as the cases. They will be randomly sampled from a general roster of the residents.

Ethics and dissemination

This study was approved by the Bioethics Committee of the Medical Association of Uzbekistan (approval report of meeting #1 on 18 January 2024). The results of this study will be widely disseminated at scientific meetings and conferences, and published in a peer-reviewed journal.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Tuberculosis (TB) is a highly prevalent communicable condition, only in specific areas of the world.

• The epidemiological factors of TB morbidity, in addition to Mycobacterium tuberculosis, range from low socioeconomic status to immunodeficiency or micronutrient deficiency in the community.

WHAT THIS STUDY ADDS

• It remains unclear why TB morbidity is high in the specific communities and geographical areas.

• One explanation can be the higher prevalence of the stated factors among those communities.

• Another could be the prevalence of ethnicity or/and residency-associated factors that remain unknown or not fully studied.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The finding of any ethnicity-associated or residency-associated factors affecting TB morbidity can contribute to the formulation of an effective policy on TB control in the specific area.

• This is particularly important for areas in the Aral Sea region, where the community has been suffering as a result of environmental disasters over an extended period of time.

Introduction

Tuberculosis (TB) is one of the 10 leading causes of death globally and the primary cause of death from infectious diseases, resulting in significant medical and social challenges.¹ The rate of TB-caused disability, as a result of respiratory, mental and neurological conditions, remains high in low-income countries.² A common factor is the household and outside transmission of TB, that remains high, even around the world.³ Uzbekistan is one of the 30 countries with the highest TB morbidity; together with those countries, it accounts for 87% of all cases in the world. The incidence of TB in the Republic of Karakalpakstan (RK) remains consistently high in comparison with the other regions of Uzbekistan. High indoor air pollution,⁴ with the prevalence of undiagnosed diabetes and anaemia,^{5 6} can be critical for the high rate of TB morbidity in the community. An aggravating factor in this region is the high proportion of multidrug-resistant (MDR) TB,^{7 8} with resistance acquired even to new anti-TB drugs.⁹ As an example, the resistant Beijing clusters of Mycobacterium tuberculosis became dominant in all of Central Asia.¹⁰ Previous TB treatments and social factors could be the factors contributing to the widespraed prevalence of MDR-TB in RK.¹¹

The highest rates of TB morbidity and mortality did not change in RK during the antibiotic era of TB control. Some exception to this trend was observed in the 1980s when the incidence rates became similar to those of the rest of Uzbekistan. One possible reason for such a fall in TB morbidity in RK could be the broad introduction of anti-TB treatment or even a natural scenario for this endemic condition.^{12 13} Traditionally, a higher burden to health in this region is explained by the ecological disaster caused by the draining of the Aral Sea. However, data showed that the TB incidence rate in RK was twice that of the rest of Uzbekistan even in 1953, that is, the during the time when the disaster had not entered its initial stage. Moreover, the incidence rates of TB strongly differ even among the regions inside RK. Data for 2019 showed that the highest rates (200–300 per 100 000) were registered in the northern and north-western regions, the lowest (40–70 per 100 000) in the southern regions, with intermediate rates registered in the central regions. In addition, a spatial difference was seen in the effectiveness of anti-TB treatment¹ among the regions. An important question is the extent to which specific residencies contribute to TB incidence at the individual level. If it is proven to be a contributor, it becomes important to discover immediate epidemiological factors of high TB morbidity in the community in RK. These factors may still be unknown or are underestimated in public health policy.

The ethnic composition of the community in RK can be one of the possible causes for such geographical heterogeneity in TB morbidity here. This means that some factors can be ethnicity-associated determinants related to communities living in this area. Historically, people of Uzbek ethnicity have lived predominantly in the southern regions of RK and have followed a sedentary lifestyle for a long time, being mainly engaged in agriculture. People of Karakalpak and Kazakh ethnicity lived in the northern and north-western regions, and followed a seminomadic and nomadic lifestyle until the 1940s. These regions became the ones with the highest TB incidence in RK. However, the probability of contact with a source of TB infection is logically lower in the communities that follow a seminomadic or nomadic lifestyle than in a community living a sedentary life. A review of TB incidence among UK ethnic minorities¹⁴ concluded that factors relating to M. tuberculosis transmission or social factors may potentially affect ethnic groups and contribute to differences in TB incidence. However, this also means that not all factors related to TB incidence are ethnicity-associated in the specific geographical area. Poverty, overcrowding or HIV have been seen to be factors of TB mortality in England and Wales for 10 years; however, none of them were related to ethnicity.¹⁵ A mutation in the interferon-gamma gene is a well-established genetic factor influencing the risk of TB. An ethnicity-based analysis did not confirm such an association for a population with Asian ethnicity but did so for East Asian, African and American populations.¹⁶ A meta-analysis¹⁷ proved that vitamin D deficiency is a risk factor for active TB and, previously, vitamin D receptor gene methylation was found to be an ethnically associated factor for its deficiency.¹⁸ Ethnicity-associated transmission of M. tuberculosis was found in the Netherlands,¹⁹ where higher transmission indices were seen in the non-Dutch populations. In case ethnicity is found to be a risk factor for TB morbidity or mortality in RK, such characteristics become factors that impact differently on the ethnic groups under comparison, and finding such factors will contribute to targeting them in the national policy.

Methods and analysis

Study objectives

An evaluation of the ethnic composition, as an independent factor of TB incidence, has never been performed inside or outside the framework of registry-based cohort studies in Uzbekistan or RK. The reason for this could be that data on the ethnicity of patients with TB were either not introduced to the standard registries or not considered a variable for study. The rationale for both residency and ethnicity evaluation, as independent factors under the control of known factors, is that in case: (A) We prove it, ethnicity/residency-related factors will be other than the known factors and (b) We do not prove it, ethnicity/residency will be associated with the known factors. In both scenarios, we get important data for prospective research and public health policy in the region. So, we have two principal aims of our study:

Aim#1: To evaluate the role of ethnicity and residency on the development of active TB in the community living in RK.

Hypothesis #1a: Individuals of Karakalpak/Kazakh ethnicity will have a higher risk of developing active TB than individuals of other ethnicities.

Hypothesis #2b: Individuals living in the northern and north-western regions of RK will have a higher risk of developing active TB than individuals living in other regions.

Aim#2: To evaluate the association between factors contributing to the development of active TB and ethnicity/residency, in the community living in RK.

This is the secondary aim of the study, and to address it, we will match the statistical data collected to test the hypotheses under aim#1. For this reason, the list of variables and related data will be substantially extended to have the necessary data for the evaluation. We do not formulate a specific hypothesis towards this aim at the planning stage, as it will depend on the results of the preliminary analysis.

Study design

This study has a comparative case-control observational design to test the hypotheses under its aims. This type of design is considered appropriate for a rare condition such as TB. A retrospective approach will contribute to collecting the necessary data with a smaller sample size. The most critical type of bias in the comparative study becomes the one caused by confounders whose effects are not studied in the research.²⁰ To control such factors, the confounders will be equally distributed between the groups under comparison from the beginning of the study or they will be matched in groups at the analysis stage. Even though study hypotheses relate only to ethnicity/residency, the case and control groups will be matched only by sex and age for evaluation of the role of additional factors. We believe this approach provides better use of the study resources. An additional issue is that we prevent ‘overmatching’ which can restrict the range of the evaluated factors.²¹ For better control of confounders, we plan to engage controls and cases in a 1:1 ratio. Summarised descriptions of the groups with their estimated sample sizes and data sources are shown in table 1.

Table 1. Study groups, data sources and sample sizes.

Groups	Description	Data sources	Sample size
Cases	Includes randomly sampled subjects with clinically diagnosed or bacteriologically confirmed TB	Registry of patients from RK with TB diagnosed during 2016–2020	911
Controls	Includes randomly sampled subjects without TB	Roster of residents from RK without TB, during 2016–2020	911

RKRepublic of KarakalpakstanTBtuberculosis

Case definition and inclusion criteria

We referred to the WHO’s ‘Definitions and reporting framework for tuberculosis – 2013 revision (updated December 2014 and January 2020)’, to define cases for both the case and the control groups.

The case group of the study will include patients with clinically diagnosed TB that are based on X-ray abnormality or patients with TB bacteriologically confirmed by the GeneXpert MTB/RIF assay.

The control (healthy) group will include individuals having no TB both at the time of examination and historically. This means that individuals with current or past TB will be excluded from the control group. To guarantee the inclusion of healthy individuals, all selected individuals will be examined by a specialist (pulmonologist) and tested by the X-ray method. History of TB will be checked via medical records and a registry of patients with TB. Individuals with latent TB, that is, those having TB without symptoms, will be excluded from both the case and the control groups of the study.

Study settings

Cases will be randomly sampled from the registry of patients with TB from RK, who were diagnosed with TB (IСD-10: A15–A16) during 2016–2020. Controls will comprise individuals without TB, living in RK at the same time as the cases. They will be randomly sampled from a general roster of the residents. This means that the study will potentially include any of the residents of RK. This approach is important because residency becomes the evaluated variable. Selecting controls from the same population will provide the same base conditions as the case group.²² This will allow inclusion of subjects with and without TB from different regions of RK for evaluation of the role played by a specific ethnicity/residency.

Study outcomes

As a primary outcome, we will consider the incidence of active TB in the studied population, that is, residents of RK. We will investigate the role of ethnicity and residency-related variables at the individual level, to test the hypotheses under the aim#1. We will evaluate the OR between the cases and controls, and this ratio becomes crucial for the final conclusion and for the calculation of the sample size of the study. We will adjust the ratio to potential confounders such as socioeconomic status, immunodeficiency (HIV, diabetes) and malnutrition (anaemia). These will be the secondary outcomes to be investigated in our study. Analysis of residency data will consider the distance from the Aral Sea as a possible environment-related factor contributing to TB incidence. Ecological-level data analysis will consider the TB morbidity indices in different communities of RK.

Sample size calculation

We considered a minimal (95%) CI, statistical power (80%), an expected OR (1.5) and a per cent of controls exposed (10%) as the initial criteria for calculation of the sample size. Considering previous data on at least double difference in TB morbidity among regions of RK, we could accept an OR of at least 1.5 to test hypotheses under aim#1. However, we believed this would be an unacceptable assumption for the additionally tested exposing factors. Increasing power to 90% increases the required sample size to 2500. Another important rate expected to affect the sample size is the planned ratio of controls to cases. To control potential confounders, we decided to include one control for every included case. It is evident, however, that this assumption could result in uncertainties. Finally, we opted to aim for an OR of 1.5 and target at least 911 ‘case’ and 911 ‘control’ subjects.

Data analysis

The analysis will be based on data on case-control study subgroups, and we will estimate ORs with 95% CIs for the distribution of measured risk factors (ethnicity/residency and additional factors). We will apply the GIS tool to point out residency and migration locations for the subjects and their declared ancestors in both groups. Associations of TB incidence and specific ethnicity or residency with the binary outcomes will be assessed using generalised linear models with a Poisson distribution and log link function. We will set statistical significance at p<0.05 for all statistical tests. In all regression models, we will adjust first for potential sociodemographic confounders, such as employment, disability, diabetes, malnutrition (anaemia) and HIV status. We will also apply the χ² distribution test with analysis of frequencies of categorical variables and the Mann-Whitney U test for continuous variables. The Cochran-Mantel-Haenszel method will be used for stratified analysis to statistically control confounders.²³

Data collection and research schedule

Study participants in the two groups will be recruited from the community living in RK. First, for the case group, we will include subjects from the registry of patients ≥18 years old with clinically diagnosed TB during 2016–2020. We will attempt to get primary data (residency or ethnicity) on patients from the registry. However, we consider receiving some information, such as detailed ethnicity or income data, from the participants. This means that we would contact them by phone or personally, for an interview. Anyway, we would get their consent by phone for contact with the research team. In case they agree, we will collect data by phone or meet them personally. For the control group, we will recruit healthy (non-TB) subjects from the roster of residents in RK. We will contact them in the same way as the case group participants.

We will use specific tools for data collection and for interviewing the study subjects. As we prefer to expand the number of measurable variables, we will develop data forms and questionnaires for their collection. However, we prefer to use medical records and registry as a data source, considering the high chance of recall or unwillingness to open issues, especially concerning health conditions (HIV, diabetes). For the case group, data will be collected from the registry and inpatient/outpatient records. For the control group, data will be collected from primary care (outpatient) medical records. The schedule of data collection with the study timeline is provided in figure 1. The whole period of study preparation with data collection, analysis and publication is expected to be 24 months. The starting point for recruitment is 1 February 2024 when data were accessed for the research. We expect to complete the recruitment and make the entire data set available for analysis by the end of 2025.

Figure 1. Research schedule and description of the steps (Q1–Q8). At present, the preparation under Q1–Q2 is completed and data collection has started. Data collection is the most time-consuming step, especially for the control group. The Republic of Karakalpakstan (RK) has a large area, sometimes with severe conditions in the winter and summer. The principal investigator will report to the funding agency in case additional time is needed for each step. TB, tuberculosis.

Safety considerations

A possible safety concern expected during the interview process is close contact between the interviewer and the study subject. For safety of both the subject and interviewer, they will use face masks (preferably N95) with standard hand sanitisers to prevent both TB and COVID-19 infection. A standard COVID-19 vaccination is provided under the country’s (Uzbekistan) policy.

Ethics and dissemination

This planned study has a retrospective comparative design and will include human subjects. However, our study does not match the definition of clinical trial accepted, for example, by the US National Institutes of Health.²⁴ The main point is that there is no possibility of any intervention in the natural flow of any condition on involved subjects, under the retrospective design. The study just observes data on subjects concerning relations between the studied outcomes and exposure. We will not interfere clinical practice of caring TB patients. From this view, the individual risk for participants, in our study, becomes minimised or zero level.

Informed consent and confidentiality

To provide autonomy to the subjects, all the subjects in both the study groups will sign an informed consent form, according to the requirement of WMA declaration of Helsinki.²⁵ The informed consent form will be provided and filled in a language (Karakalpak, Uzbek and Russian) selected by the subjects. Since a significant portion of the information will be collected retrospectively from registries, medical records and national databases, study participants will provide their consent for the use of these data in the research. In case the participant is incapable of providing such consent, his relatives are allowed to do so. Patients who have died by the time of data collection or were in severe condition due to TB or other causes, would be excluded from the study. In any case, researchers accept an obligation not to disseminate subjects’ personal information to third parties or other persons. Standard data collection forms will not include personal information (name, residency) about the subjects to prevent any possibility of use of this information by third parties or persons.

Other issues

An issue that can possibly arise at the end of the study concerns its interim conclusion that may stress the association of an active TB with some ethnicities. Such a conclusion may cause a discriminatory view relating to this ethnicity. Ethnicity is rather a widely studied characteristic in epidemiology, and we declare that we do not have an objective to discriminate against any ethnicity. We also declare that none of the results can be used for a discriminatory goal. However, as we accept the possibility of misinterpretation, the research team will introduce an appropriate declaration for all disseminated publications or reports.

Patient and public involvement

No patient was involved in the development of the research questions, study design, recruitment, outcome measures, dissemination plans and conduct of the study. The study findings will be disseminated through open access scientific publications, stakeholder networks and national/international meetings

Publication policy and dissemination

We consider the principal investigator to be a coauthor in all publications resulting from this research project. His name will be cited at the end of the authors’ list. All other study team participants, consultants and others will be included as coauthors depending on their scientific contribution to the specific publication. The study results will be primarily disseminated through conferences and academic courses. The results of public health policy are expected to be considered as well.

Protocol version

The current protocol version is V.4.0 (11 January 2024), which was submitted for ethical approval by the Bioethics Committee at MAU. Now it has been approved for implementation.

Discussion

Despite no consensus on what the self-reported ethnicity represents, it is a combination of genetic, socioeconomic with cultural factors and the ethnicity becomes a much-studied variable in epidemiology.²⁶ For epidemiological purposes, any group whose members lived in proximity and shared customs and values for a length of time sufficient to acquire common characteristics, whether by biological or social mechanisms, is of interest. We consider limitations, specific to case-control study design and relating a high chance of unmeasured confounding in such studies.²⁰ Another issue is the memory (recall) bias on subjects in retrospective studies. To prevent it, all possible ranges of information concerning subjects, except their cultural and ethnicity data, will be collected from medical records.