Discriminatory attitudes toward people living with HIV/AIDS in East Africa: a multilevel analysis of socioeconomic and behavioral determinants using DHS 2019–2023 data

Abstract

Introduction

Despite global efforts to reduce HIV-related stigma, discriminatory attitudes toward people living with HIV/AIDS (PLHIV) remain widespread, especially in Sub-Saharan Africa. This study aimed to assess the prevalence and determinants of discriminatory attitudes toward PLHIV in five East African countries using Demographic and Health Survey (DHS) data from 2019 to 2023.

Methods

A cross-sectional, secondary data analysis was conducted using DHS data from Kenya, Madagascar, Mozambique, Rwanda, and Tanzania. A total weighted sample of 106,639 sexually active individuals aged 15–49 was included. Discriminatory attitudes were assessed based on responses to two standard DHS questions. Multilevel logistic regression models were used to identify factors associated with discriminatory attitudes at both individual and community levels.

Results

The overall prevalence of discriminatory attitudes toward PLHIV was 34.03% (95% CI 33.74, 34.31). Country-specific prevalence ranged from 12.98% in Rwanda to 76.18% in Madagascar. Key factors significantly associated with discriminatory attitudes included lower wealth status (AOR = 2.12, CI 1.73, 3.80), having multiple sexual partners (AOR = 2.11, CI 1.97, 2.60), and being unemployed (AOR = 3.10, CI 2.06, 3.70).

Conclusion

Discriminatory attitudes toward PLHIV remain high in East Africa, particularly among economically disadvantaged, unemployed individuals and have multiple sexual partners. Integrated interventions targeting socio-economic inequalities, sexual health education, and employment opportunities help reduce discriminatory attitudes toward PLHIV.

Introduction

Human immunodeficiency virus (HIV) remains a major global public health concern, with Sub-Saharan Africa carrying a disproportionately high burden [1, 2]. Stigma and discrimination toward people living with HIV/AIDS (PLHIV) are widely documented, especially among individuals who are already marginalized due to socioeconomic status, gender, or race. Such stigma continues to deter millions from seeking testing, prevention, and treatment services [3, 4].

People living with HIV (PLHIV) may encounter three primary forms of stigma: self-stigma, or negative self-perception; anticipated stigma, the expectation of discrimination upon disclosure; and enacted stigma, which refers to actual discriminatory behaviors [5].

Community members also demonstrated discriminatory behaviors by avoiding sitting near people living with HIV (PLHIV), moving away from chairs adjacent to them, and acting as though they were reluctant to have direct physical contact with them [6]. In order to understand how uninfected people react to those who are infected, stigma primarily takes the form of prejudice, stereotyping, and discrimination against PLHIV. The term “prejudice” describes the negative attitudes that non-infected people have about PLHIV, such as disgust, rage, and dread [6, 7].

Sub-Saharan Africa continues to bear the highest global HIV burden. In particular, Eastern Africa remains one of the most affected regions, with a substantial proportion of people living with HIV and new infections occurring there. Although several countries in the region have shown recent declines or stabilization in HIV prevalence, the overall burden remains high [8].The HIV epidemic remains most severe in Eastern and Southern Africa, which together account for a large proportion of global HIV cases and new infections. Beyond its medical impact, HIV/AIDS is also a major social issue. People living with HIV frequently experience stigma that affects their rights, access to healthcare, and social relationships. Such stigma can hinder timely diagnosis and treatment, reinforcing social isolation and contributing to ongoing transmission [9, 10]. Stigma can result in discrimination along with other civil rights breaches that have a profoundly negative impact on the welfare of people who live with HIV [7].

One of the stated barriers to achieving universal access to HIV/AIDS prevention, treatment, care, and support services is discrimination, which is defined as a distinct action or behavior toward the stigmatized individual based on those beliefs and perceptions [11]. The nature of HIV/AIDS, which includes its mortality and incurability, contagiousness and transmissibility, and the disease’s repulsive, unsightly, and distressing appearance in its mature stages, is the cause of discrimination associated to the disease [11, 12].

At the community and family levels, stigma and discrimination can result in delayed care-seeking, medication non-adherence, psychological distress, verbal and physical abuse, lack of social support, and risky health behaviors such as hiding medications [13, 14]. Furthermore, stigma and discrimination are held accountable as major barriers to the effectiveness of HIV preventive and care initiatives [15]. PLHIV, their families, and the medical professionals who treat them are all significantly impacted by discrimination [10, 16, 17].

Previous research identifies multiple factors associated with discriminatory attitudes toward PLHIV, including education, financial and employment status, residence, media exposure, HIV testing history, marital status, region, high-risk behaviors, disease stage, and personal relationships with someone living with HIV [15, 18]. In low- and middle-income countries, additional factors such as older age, poverty, lack of HIV awareness, unsafe sexual practices, and possession of health insurance have been implicated [19–21]. A thorough understanding of HIV, educational attainment, exposure to the media, economic standing, and having ever been tested were all linked to the prejudiced attitudes of those living with the virus in low- and model-income nations [18, 22, 23].

Given the significant public health implications, there need for effective policies and programs to reduce discriminatory attitudes and support individuals affected by HIV/AIDS. Despite well-documented risks, few studies have examined the burden and determinants of discriminatory attitudes at both the individual and community levels using nationally representative datasets. In this study, discriminatory attitudes are operationalized using standard Demographic and Health Survey (DHS) proxy indicators that primarily reflect social distance and exclusion rather than internalized stigma or enacted discrimination. Therefore, this study aims to determine the prevalence and determinants of discriminatory attitudes toward people living with HIV/AIDS in East Africa using the most recent DHS data from 2019 to 2023.

Method

Data source and population

We conducted a multilevel analysis to examine discriminatory attitudes toward people living with HIV/AIDS and their associated factors among sexually active men and women aged 15–49 years (commonly defined as reproductive age) in East Africa. Data were drawn from the most recent Demographic and Health Surveys (DHS) conducted between 2019 and 2023 in five East countries: Kenya, Madagascar, Mozambique, Rwanda, and Tanzania. The DHS is a nationally representative, community-based cross-sectional survey providing information on health and related indicators. Our study focused on sexually active individuals aged 15–49 years, yielding a final weighted sample of 106,639 participants. Further details on the DHS methodology are available website. https://dhsprogram.com/Methodology/index.cfm.

Variables of study

Outcome variable

Discriminatory attitudes toward people living with HIV were measured using two standard Demographic and Health Survey (DHS) items assessing social distance and exclusion: [1] whether children living with HIV should be allowed to attend school with children without HIV, and [2] whether respondents would purchase vegetables from a vendor living with HIV. Participants who answered “no” to either item were classified as having discriminatory attitudes. These two DHS items have been widely used as proxy indicators of discriminatory attitudes toward people living with HIV/AIDS in population-based studies across East Africa. While commonly applied in DHS-based analyses, these items primarily capture social distance and social exclusion and do not provide a comprehensive assessment of HIV-related stigma, such as internalized stigma, enacted discrimination, or structural stigma.

Independent variables

Both individual and community-level factors were used to measure discriminatory attitudes against people living with HIV/AIDS. Individual-level factors were age, occupation, education, health visits, household headship, marital status, economic status, history of HIV testing, multiple sexual partners, cigarette smoking, and history of sexually transmitted diseases were among the independent variables that were taken from the DHS data of the five East African countries (2019–2023). The community-level determinants of discriminatory attitudes toward people living with HIV/AIDS in East African countries included residence (urban vs. rural) and media exposure (internet use: low vs. high). These variables were constructed by aggregating individual responses within each cluster using Stata, with cluster proportions calculated in Excel. For normally distributed variables, the mean was used, and for skewed variables, the median was applied. Clusters were then categorized based on these values. These variables were included in the multilevel logistic regression models to examine their association with discriminatory attitudes toward PLHIV.

Statistical analysis

Data were extracted from the DHS datasets and managed using Stata version 14. Variables were recoded and cleaned prior to analysis to ensure consistency across countries. Descriptive statistics were computed to summarize the study variables and are presented as frequencies and percentages. Sampling weights were applied to account for the complex survey design and non-proportional allocation, ensuring national representativeness were performed using sample weight with cluster. To maintain the hierarchical structure of the gathered data, a mixed multilevel analysis was performed.

Multilevel bivariable logistic regression was first performed to identify candidate variables for inclusion in the multivariable analysis using a p-value threshold of < 0.25. Subsequently, multilevel multivariable logistic regression was conducted to determine statistically significant associations at p < 0.05. Four models were fitted: a null model without explanatory variables, a model including only individual-level variables, a model including only community-level variables, and a final model incorporating both individual- and community-level variables. Adjusted odds ratios (AORs) with 95% confidence intervals (CIs) were reported and sampling heights were applied to account for the complex survey design and ensure representativeness.

The proportional change in variance (PCV) was calculated as PCV = (Vnull​−VA)/Vnull​×100% where VA represents the cluster-level variance and Vnull the variance in the null model. Furthermore, the degree of heterogeneity in discriminatory attitude between the clusters was assessed using the intra-class correlation coefficient (ICC), which is calculated using the formula ICC = VA/VA + 3.29∗100%. This coefficient measures the percentage of the total observed individual variance in discriminatory attitude that can be attributed to differences between clusters. The median odds ratio (MOR) was used to quantify the variation in discriminatory variables across clusters. MOR = e^0.95√VA [24]. Finally, the adjusted odds ratio (AOR) with a 95% confidence interval (CI) was calculated, and variables statistically significantly associated with discriminatory attitudes were identified with a p-value of less than 0.05.

Result

Socio-demographic characteristics of the respondents

A total of 106,639 participants from east African nations were included in this study. Approximately 23,551 (22.08%) of the study participants were in the 15–19 age range and 67,449(63.25%) were females. 47,554 (36.50%) of the survey participants had a secondary education. 69,651 people (65.31%) lived in rural areas. 40,749 (38.21%) of the study participants were married, whereas 82,435 (77.30%) had no sexual partner. 74,317 (69.79%) of the study participants were employed. 68,931 (64.64%) of the study’s participants had an HIV test at some point in their life (Table 1).

Table 1.

Socio-Demographic characteristics of study participants from the demographic and health surveys people living with HIV/AIDS (PLHIV) in five East African countries (n = 106,639)

Individual level variables	Category	Frequency (n)	Percent (%)
Age of participant	15–19	23,551	22.08
	20–24	19,250	18.05
	25–29	16,674	15.64
	30–34	14,339	13.45
	35–39	13,274	12.45
	40–44	10,668	10.00
	45–49	8883	8.33
Education level of participants	No education	12,150	11.39
	Primary education	47,554	44.59
	Secondary education	38,922	36.50
	Higher education	8013	7.51
Residence of participants	Urban	36,988	34.69
	Rural	69,651	65.31
Working status	Have no working	32,173	30.21
	Have working	74,317	69.79
Number of sexual partner	No sexual partner	82,435	77.30
	One sexual partner	19,122	17.93
	Multiple sexual partner	5082	4.77
Currently smokes tobacco	Yes	5021	4.71
	No	101,618	95.29
Wealth index	Poor	35,486	33.28
	Middle income	20,877	19.58
	Rich	50,276	47.15
Having sexual transmitted disease	Yes	4270	4.00
	No	102,369	96.00
Being tested	Yes	68,931	64.64
	No	37,708	35.36
Usually health visit	Yes	75,057	70.38
	No	31,582	29.62
Internet use of participants	Yes	31,073	29.14
	No	75,566	70.86
House head participants	Male	77,666	72.83
	Female	28,973	27.17
Marital status	Never in union	27,311	25.60
	Married	40,749	38.21
	Living with partner	22,584	21.20
	Separate/widowed	15,995	14.99

Prevalence of discriminatory attitude toward people living HIV/AIDS (PLHIV)

Analysis of the Demographic and Health Surveys conducted in five East African countries between 2019 and 2023 showed that 34.03% of respondents (95% CI: 33.74, 34.31) reported discriminatory attitudes toward people living with HIV/AIDS. Cross-country variation was evident, with Madagascar exhibiting the highest prevalence, Rwanda the lowest, and Mozambique, Tanzania, and Kenya showing intermediate levels (Fig.1).

Fig. 1.

Prevalence of discriminatory attitudes towards people living with HIV/AIDS (PLHIV) in each of the Five East African Countries 

Random effect model and model fitness

The ICC value in Model 1 shows that, as seen in (Table 2), differences between communities or clusters account for about 49.95% of the diversity in discriminatory attitudes toward People Living with HIV/AIDS (PLHIV). The highest PCV value in Model 3 also shows that both individual and community-level factors account for 30.06% of the variation in discriminatory attitudes toward individuals living with HIV/AIDS. Additionally, Model 3’s lowest deviation of 1405.33 suggests that it fits the data the best.

Table 2.

Random effects and model fit statistics for discriminatory attitudes toward PLHIV in five East African countries (n = 106,639)

Parameter	Null model	Model I	Model Ⅱ	Model Ⅲ
Variance	4.340518	3.154586	3.084707	3.035403
ICC	58.88%	48.95%	48.39%	47.99%
MOR	7.24	5.40	5.30	5.23
PCV	Reference	27.32%	28.93%	30.06%
LLR	​−732.93272	​−725.84015	​−711.74356	​−702.66417
Deviance	1465.86	1451.68	1423.49	1405.33

Factors associated with discriminated attitude people living with HIV/AIDS

In the multilevel bivariable logistic regression analysis, individual-level factors including working status, educational level, wealth status, and having multiple sexual partners were associated with discriminatory attitudes toward people living with HIV/AIDS. In the multilevel multivariable logistic regression analysis, poor wealth status, multiple sexual partners, and unemployment remained significantly associated with discriminatory attitudes. Participants with poor wealth status had higher odds of holding discriminatory attitudes compared with those in the middle and highest wealth categories (AOR = 2.12, CI 1.73, 3.80). Respondents reporting multiple sexual partners also had higher odds of discriminatory attitudes compared with those reporting one or no sexual partners (AOR = 2.11, CI 1.97, 2.60). Similarly, unemployed participants had higher odds of discriminatory attitudes than employed participants (AOR = 3.10, CI 2.06, 3.70) (Table 3).

Table 3.

Multilevel multivariable logistic regression analysis in five East African countries (n = 106,639)

Individual level variables	Category	Model Ⅰ AOR(95% CI)	Model Ⅱ AOR(95% CI)	Model Ⅲ AOR(95% CI)	P-value
Age of participants	15–19	0.61 (0.52, 1.69)		1.83 (0.68, 2.00)	0.072
	20–24	0.62 (0.54, 1.70)		1.90 (0.76, 2.05)	0.058
	25–29	0.48 (0.41 1.56)		1.65 (0.52, 1.78)	0.064
	30–34	0.26 (0.19, 1.34)		1.32 (0.22, 1.43)	0.090
	35–39	0.16 (0.09, 1.24)		1.20 (0.10, 1.29)	0.070
	40–44	0.09 (0.01, 1.16)		1.10 (0.01, 1.19)	0.056
	45–49	1		1
Marital status of participants	Single	1		1
	Married	1.10 (0.03, 1.16)		2.99 (0.81, 3.19)	0.074
	Living with partner	0.71 (0.64, 1.78)		2.01 (0.88, 2.16)	0.057
	Separated/widowed	0.71 (0.63, 1.78)		1.99 (0.85 2.15)	0.076
House head	Male	0.059 (0.01, 1.10)		1.05 (0.01, 1.10)	0.089
	Female	1		1
Health visit	Yes	0.21 (0.15, 1.27)		0.82 (0.79 1.85)	0.078
	No	1		1
Being tested	Yes	0.72 (0.66, 1.78)		0.18 (0.174 1.19)	0.087
	No	1		1
Having sexual transmitted disease	Yes	0.30 (0.19, 1.36)		0.74 (0.67, 2.81)	0.075
	No	1		1
Wealth status	Rich	1		1
	Middle income	0.17 (0.11, 1.22)		0.84 (0.80 1.88)	0.067
	Poor	1.07 (1.21, 2.33)		2.12 (1.73 3.80)	0.010
Currently smokes tobacco	Yes	0.71 (0.46, 1.03)		0.49 (0.24, 1.05)	0.063
	No	1		1
Having sexual partner	No sexual partner	1		1
	One sexual partner	0.37 (0.32, 1.43)		0.68 (0.64 1.71)	0.058
	Multiple sexual partner	1.46 (1.33, 1.59)		2.11 (1.97 2.60)	0.01
Working status	Yes	1		1
	No	1.09 (1.06, 1.43)		3.10 (2.06, 3.70)	0.02
Level of education	No education	1.64 (1.29, 1.89)		0.53 (0.50, 1.56)	0.072
	Primary education	0.86 (0.72, 1.90)		0.45 (0.43, 1.48)	0.056
	Secondary education	0.49 (0.39, 1.58)		0.28 (0.25 1.31)	0.091
	Higher education	1		1
Community level variables
Internet use	Yes		0.96 (0.92 1.99)	0.72 (0.68, 1.76)	0.063
	No		1	1
Living residence	Urban		1	1
	Rural		1.19 (0.16 1.22)	0.53 (0.50, 1.56)	0.074

Note: Boldface indicates statistically significant variables (p>0.05).

Discussion

This study found that discriminatory attitudes toward people living with HIV/AIDS remain common across East African countries and are associated with key individual- and community-level factors. This multilevel analysis indicated that poor wealth status, unemployment, and having multiple sexual partners were significantly associated with higher odds of discriminatory attitudes, while community-level factors accounted for a substantial portion of the variation. Interpreted through structural and behavioral HIV frameworks, limited economic resources and unemployment may reduce access to information and social networks, while multiple sexual partners may influence perceived risk and social judgment, shaping stigmatizing attitudes [7, 25, 26].

These findings suggest that Discriminatory attitudes toward PLHIV reduction efforts in East Africa should address socioeconomic conditions and individual risk-related factors. Approaches that improve economic stability and employment may help mitigate social conditions associated with discriminatory attitudes, while accurate messaging around sexual risk could counter moralized perceptions linked to multiple sexual partnerships. Addressing these interconnected factors may support more comprehensive stigma reduction strategies in the region.

This finding is lower than other studies conducted in Chinese (42%) [27] Indonesia [28], Africa countries which is Ghana DHS (60.92%) [29], sub- Saharan African countries DHS (47.08%) [22], Ethiopian country DHS (62.66%) [23]. The disparity in attitude could be Discrimination has lessened in East African countries as a result of long-running, comprehensive initiatives to increase knowledge of HIV and lessen stigma. Another possibility is that more tolerant attitudes toward people with HIV were influenced by cultural and religious factors as well as increased access to antiretroviral therapy. But in Ghana, Ethiopia, and Indonesia, for example, stigma can still be more firmly embedded because of gender dynamics, cultural norms, and different transmission patterns. In studies that solely focus on women, who often face stigma that is made worse by their gender and HIV status, this is especially true. These discrepancies are also influenced by regional socioeconomic factors and differences in study design [27, 30, 31].

In others this study is greater than studies conducted in East Africa countries among only women DHS data 32.73% [26], among women in three sub-Saharan African countries 28.19% [32]. This disparity may be explained by variations in the study populations, the timing of data collection, and the techniques used to assess the pertinent result. Contextual elements including cultural customs, knowledge levels, healthcare access, and socioeconomic status disparities may also have an impact on the apparent inequality. Studies show that such contextual and methodological differences can significantly affect prevalence estimates [22, 33].

Regarding factors associated with discriminatory attitudes toward people living with HIV, participants who are unemployed are three times more likely to exhibit discriminatory attitudes towards people living with HIV compared to those who are employed. This result aligned with the findings of earlier investigations [34]. This could be explained by the fact that unemployment is often associated with lower levels of education, less access to health information, and less exposure to awareness campaigns all of which are critical in forming attitudes that are not stigmatizing. Furthermore, because jobless people frequently face social marginalization and financial difficulty, they may be more susceptible to adopting biased viewpoints as a coping strategy or as a distraction from their own financial difficulties. According to research, having a job not only improves socioeconomic standing but also increases social interaction and educational opportunities, all of which lead to more knowledgeable and compassionate views of those living with HIV [35, 36].

Another factor is that participants who have multiple sexual partners are twice as likely to exhibit discriminatory attitudes toward people living with HIV compared to those who have only one or no sexual partner. Other studies on women in East African nations were linked to this association [26].The reason behind this could be that those who have more sexual partners may believe they are more likely to spread HIV, which increases stigma and dread of people who are HIV positive. These people might also have false assumptions about how HIV spreads or link it to promiscuity, which would fuel their prejudice. Studies have indicated that because of their ignorance or fear of the virus, persons who participate in higher-risk sexual practices frequently grow more hostile toward those who are HIV positive [35, 37, 38].

Additionally, participants from lower wealth status households are twice as likely to exhibit discriminatory attitudes toward people living with HIV compared to those from middle or high wealth status households. This result aligned with the findings of earlier investigations in 15 sub-Saharan African [22]. Numerous factors could be responsible for this connection. People from lower socioeconomic backgrounds frequently have less access to healthcare, education, and HIV knowledge, which can result in stigmatization and misinformation. Increased stress and social exclusion brought on by economic difficulties can also intensify negative views toward other people as a coping strategy. Additionally, research has demonstrated that in Ethiopia, people from lower-income households are less likely to get tested for HIV, have thorough understanding of HIV/AIDS, or have positive views toward those who are HIV positive. These differences show how critical it is to address socioeconomic gaps in order to lessen stigma and discrimination associated with HIV [39–41].

Strength and limitation of study

This study benefits from a large, nationally representative 2019–2023 Demographic Health Survey (DHS) dataset across five East African countries, offering valuable insights into individual and community-level factors affecting discriminatory attitudes toward people living with HIV/AIDS. The multilevel analysis and large sample size of 106,639 participants enhance the reliability of the results. However, limitations include reliance on secondary data, which limits the exploration of additional factors, and the cross-sectional nature, which prevents causal inferences. Additionally, this study relied on two DHS proxy indicators of discriminatory attitudes, which capture social distance and exclusion but these items do not provide a comprehensive assessment of HIV-related stigma, including internalized, enacted, or structural stigma. The restriction of the analysis to individuals aged 15–49 may limit the representation of other populations affected Discriminatory attitudes, while cultural and policy differences across countries may affect the generalizability of the findings. Furthermore, reliance on self-reported data may introduce reporting bias or measurement inaccuracies.

Conclusion and recommendation

In five East African nations, the study used DHS data from 2019 to 2023 to determine the high prevalence (34.03%) of discriminatory attitudes toward individuals living with HIV/AIDS. Unemployment, having several sexual partners, and a lower wealth level were important factors linked to these attitudes. The enduring impact of behavioral variables and socioeconomic disadvantage on HIV-related stigma is underscored by these findings. Initiatives for economic empowerment, focused awareness campaigns, and thorough sexuality education should be given top priority in order to address this, especially for low-income and jobless groups. Further measures to promote more inclusive attitudes and enhance health outcomes for people living with HIV include incorporating HIV information into job training programs and bolstering health systems’ efforts to reduce stigma.

Abbreviations

AIC

Akaike Information Criteria

AOR

Adjusted odd ratio

DHS

Demographic Health Data

CI

Confidence interval

ICC

Intra-class correlation

MOR

Median odds ratio

PCV

Proportional change in variance

PLHIV

People living with HIV/AIDS

Author contributions

GK: Writing – original draft, Writing – review & editing. GT: Writing – original draft, Writing – review & editing. GN: Writing – original draft, Writing – review & editing. EG: Writing & editing. SMB: Writing – review & editing. FA: Writing – original draft, Writing – review & editing. YW: Writing – original draft, Writing – review & editing. BMW: Writing – original draft, Writing – review & editing. SF: Writing – original draft, Writing – review & editing. MK: Writing – original draft, Writing. TS: Writing – original draft, Writing – review & editing. GR: draft, Writing – review & editing and MM: Writing – original draft, Writing – review & editing.

Funding

No funding was obtained for the research, writing, or publication of this article, according to the author(s).

Data availability

Online repositories contain the datasets used in this investigation. The article or additional material contains the names of the repository or repositories and the accession number or numbers.

Declarations

Ethics approval and consent to participate

The Demographic and Health Survey (DHS) data used in this study from Kenya, Madagascar, Mozambique, Rwanda, and Tanzania were collected in accordance with both national and international ethical guidelines. Ethical approval was granted by ICF International and the Institutional Review Board of the U.S. Centers for Disease Control and Prevention (CDC). At the time of data collection, informed consent was obtained from all participants. For individuals under the age of 18, consent was obtained from a parent or legal guardian, and assent was obtained from the minors, following standard DHS ethical procedures. The purpose of the study and the voluntary nature of participation were clearly explained to all participants by trained data collectors. Data were collected anonymously, and no personally identifiable information was used in this analysis. Ethical guidelines were strictly followed to ensure participant privacy and confidentiality, with minimal risk involved. Further information on DHS data collection procedures and ethical standards is available at www.measuredhs.com.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.