Introduction: Resource-constrained countries (RCCs) have the highest burden of cervical cancer (CC) in the world. Nonetheless, although CC can be prevented through screening for precancerous lesions, only a small proportion of women utilise screening services in RCCs. The objective of this study was to examine the magnitude of inequalities of women’s knowledge and utilisation of cervical cancer screening (CCS) services in RCCs. Methods: A total of 1,802,413 sample observations from 18 RCC’s latest national-level Demographic and Health Surveys (2008 to 2017-18) were analysed to assess wealth-related inequalities in terms of women’s knowledge and utilisation of CCS services. Regression-based decomposition analyses were applied in order to compute the contribution to the inequality disparities of the explanatory variables for women’s knowledge and utilisation of CCS services. Results: Overall, approximately 37% of women had knowledge regarding CCS services, of which, 25% belonged to the poorest quintile and approximately 49% from the richest. Twenty-nine percent of women utilised CCS services, ranging from 11% in Tajikistan, 15% in Cote d’Ivoire, 17% in Tanzania, 19% in Zimbabwe and 20% in Kenya to 96% in Colombia. Decomposition analyses determined that factors that reduced inequalities in women’s knowledge of CCS services were male-headed households (-2.24%; 95% CI:-3.10%,-1.59%; P < 0.01), currently experiencing amenorrhea (-1.37%; 95% CI:-2.37%,-1.05%; P < 0.05), having no problems accessing medical assistance (-10.00%; 95% CI:-12.65%,-4.89%; P < 0.05), being insured (-6.94%; 95% CI:-9.58%,-4.29%; P < 0.01) and having an urban place of residence (-9.76%; 95% CI:-12.59%,-5.69%; P < 0.01). Similarly, factors that diminished inequality in the utilisation of CCS services were being married (-8.23%;95% CI:-12.46%,-5.80%; P < 0.01), being unemployed (-14.16%; 95% CI:-19.23%,-8.47%; P < 0.01) and living in urban communities (-9.76%; 95% CI:-15.62%,-5.80%; P < 0.01). Conclusions: Women's knowledge and utilisation of CCS services in RCCs are unequally distributed. Significant inequalities were identified among socioeconomically deprived women in the majority of countries. There is an urgent need for culturally appropriate community-based awareness and access programs to improve the uptake of CCS services in RCCs.
The aim of this study was to examine the inequalities of women’s knowledge and utilisation of CC screening services in 18 RCCs. The point of departure of this study was to hypothesis that knowledge and screening practices of CC among women in RCCs are intricately linked to wealth. This study is the first of its kind that examines the impact of wealth on inequalities of CC screening knowledge and screening in economically poor countries. To achieve the research objective, the following three research questions (RQ) were posited: RQ 1: What is the level of women’s knowledge about CC services and the level of utilisation in RCCs? RQ 2: What are the potential factors associated with increased women’s knowledge of CC screening services and their utilisation? RQ 3: What is the magnitude of wealth inequalities in terms of women’s knowledge about CC screening services and utilisation of screening services in RCCs? This study used data from the Demographic and Health Survey (DHS) conducted across the selected RCCs. As per the study objective(s), only the latest DHS conducted in 18 RCCs were utilised [44]. The DHS is a long-standing worldwide cross-sectional household survey performed in 90 developing countries [44]. Data collection is standardised but the explored health issues vary by country. Hence, data on CC are only available for 18 RCCs. Data captured by the DHS include information on various health indicators related to maternal and child health, maternal and child mortality, fertility, family planning, nutrition, and knowledge and awareness of health, health services and health care utilisation but they vary across countries based on important local health issues. The present study was restricted in 18 resource-constrained countries (RCCs), hence, data on cervical cancer-related information are only available in these countries (Fig. 1). The DHS program collects information on knowledge, awareness and utilisation of CC screening among women from 18 resource-constrained countries only (Fig. (Fig.1):1): Albania (2017–18), Bolivia (2008), Burkina Faso (2010), Colombia (2015–16), Cote d’Ivoire (2011–12), Dominican Republic (2013), Egypt (2015), Equatorial Guinea (2014–15), Honduras (2011–12), India (2015–16), Jordan (2012), Kenya (2014), Lesotho (2014), Namibia (2013), Philippines (2013), Tajikistan (2012), Tanzania (2011–12) and Zimbabwe (2015) (Fig. (Fig.1)1) [44]. Mapping of the study settings across geographical distribution The study adopted the World Bank’s definition of resource-constrained countries (RCC), a term used to refer to all countries economically classified as low- or middle-income [45]. The RCCs are typically attributed by a lack of funds to cover health care costs, on individual or societal perspectives, which leads to limited accessibility, affroadibility, accountability and availability of healthcare services in terms of limited infrastructure, poor health systems and delivery mechanisms, and trained personnel [46–48]. Indeed, for weak health care systems, it is plausible that effects beyond women cancer may be realised and may extend to cancer more generally or to women’s health. In addition, LRCs often lack the necessary infrastructure to ensure high-quality cancer screening services and subsequent follow-up care [48]. For example, RCCs often do not have the necessary infrastructure required for ensuring high-quality cancer screening services and associated follow-up care; which in turn may be compromised by the lack of a consistent supply of both electricity, x-ray films, and technicians (engineers, technicians, and radiologists) [46]. A stratified two-stage cluster sampling is used in the most DHS surveys [49]. In the first stage, primary sampling units (PSUs) are selected from the main DHS sampling framework with probability proportional to a size measure; in the second stage, a fixed number of households (or residential dwellings) are selected from a list of households obtained in an updating operation in the selected PSUs using systematic random sampling. A PSU is usually a geographically constructed area, or a part of an area, called an enumeration area (EA), containing a number of households, created from the most recent population census. For simplicity, the DHS surveys captures two-stage surveys: the first stage is a systematic sampling with probability proportional to the EA size; the second stage is a systematic sampling of equal probability and fixed size across the EAs. This sampling procedure is usually more precise than simple random sampling at both stages. The detailed sample size calculation procedures are reported elsewhere [49], which depends on a function of the cost ratio and the intracluster correlation. where, nopt is the number of required sample, C is the total cost of the survey, c1 is the unit cost per PSU for household lising and interview, c2 is the unit cost per individual interview, n is the total number of PSUs to be selected, m is the number of individuals to be selected in each PSU, and ρ is the intracluster correlation. In this study, data from each country are nationally representative of each country’s eligible population. Eligible survey participants were surveyed through face-to-face interviews by a trained surveyor using the DHS model questionnaires. Data were collected by Measure DHS retrospectively using quantitative structural questionnaires which covered information on socio-demographic, reproductive health, access to services, and use of health services. Trained interviewers collected data via face-to-face interviews. All the data were collected at both household and individual levels of women still considered as reproductive (aged 15 to 49 years). The DHS dataset is publicly available; however, mailed consent was also taken as part of the Measure DHS protocol. Study participants were generated from the DHS as per the DHS protocol. Detailed information regarding survey sampling, quality control, management, and survey instruments are reported elsewhere [44]. Women were requested to provide information about CC screening knowledge along with awareness and utilisation of screening services. Written informed consent was taken from the respondents prior to conducting the survey. Rigorous data management was performed (e.g., data validity, reliability, quality control). This analysis considered the latest survey conducted by selected countries, and the data collection period was between 2008 and 2018. The survey response rate varied between 85 and 95%. The data set is publicly accessible after obtaining approval, which was received from the Measure DHS program. A sample was drawn from the DHS database from each of the selected RCCs. After exclusion of non-responders and participants with missing data and unusual observations, data on 1,802,413 reproductive women living in these countries were included in the analysis (Table 1). India had the highest proportion of participants, followed by Burkina Faso and the Philippines. The average age ± Standard Deviation (SD) of the participants was 35.88 years (± 7.91 SD). Distribution of study population This study considered two outcome variables, namely ‘women’s knowledge and ‘utilisation of cervical cancer screening (CCS) services’. Participants were asked knowledge-specific questions related to CC screening services [50]. More specifically, questions such as ‘have you ever heard of a pap test’, ‘Do you know what a pap test is for?’, ‘Do you know what vaginal cytology is?’, ‘Have you ever heard of vaginal cytology?’, ‘How did you learn about vaginal cytology?’, ‘In the last 12 months, have you received educational information about cervical cancer screening?’ were asked to gather knowledge-related information on CC screening. The overall women’s knowledge surrounding CC screening services was measured as a dichotomous response (1 = ‘yes’ if the participant reported any positive response about CC screening services or 0 = ‘no’ otherwise). Further, participants were asked questions related to their CC screening service utilisation; for instance, questions associated with having a pap test, gynecologic examination or vaginal cytology examination, all of which depend on available services across countries [50]. Self-reported responses for CCS screening were considered and then categorised as ‘yes’ if the participant utilised any form of CCS or otherwise ‘no’ to measure the utilisation of CCS services. Explanatory variables were selected based on the socio-ecological model for the women’s knowledge and utilisation of CCS services [40, 41], and these data were examined for potential confounders [42]. Participants’ characteristics, which included age, education, sex of the household head and age at the time of respondent’s first childbirth, were considered as the predisposing factors in the analysis. Age was grouped as follows: < 26 years, 26–35 years, 36–45 years or ≥ 46 years. Educational background was defined as no education, primary education, secondary education or higher education. Household size was classified as < 5 members, 5–7 members, and more than 8 members. Media exposure was assessed by means of access to radio and/or television, whereas health insurance coverage and wealth status were considered mediator factors. Women’s history of breastfeeding, having amenorrhea, abstaining, currently working, access to mass media exposure and having health insurance coverage were dichotomous variables (‘yes’ if present or ‘no’ otherwise). Access to medical help for the self was categorised into three groups (1 = no problem, 2 = some problem, 3 = extreme problem). SES was based on the ownership of durable assets [40]. This method has been used in previous studies employing DHS data from developing countries [39, 41, 42]. Each household’s characteristics (assets) were dichotomised (‘yes’ if present and ‘no’ if not) [51]. Country-specific principal components analysis (PCA) was performed using ownership of durable assets [40]. Weights were estimated by factor scores derived from the first principal component in the PCA. The constructed wealth index values were then assigned to individuals based on accessible variables. The wealth index was divided into five strata: poorest (Q1: lowest 20%), poorer (Q2), middle (Q3), richer (Q4) and richest (Q5: top 20%) [52, 53]. Location of residence was dichotomised as either urban or rural [52, 53]. For the inequality analysis, comparisons of knowledge CC screening and utilisation of services were performed across wealth quintiles over the period specified. The standard measures of concentration index (Conc.I) were employed to examine the magnitude of household wealth-related inequality and the trends in CC screening knowledge and utilisation of services across 18 RCCs. The Conc. I was estimated as the covariance between knowledge and utilisation of CC screening services and the proportional rank in wealth score distribution [39] as follows: where Conc. I is the concentration index, y¯ is the mean of knowledge and utilisation of CC screening services, ri is the cumulative proportion that each individual represents over the total population once the distribution of wealth score has ranked the latter. The values of Conc. I are bounded between y¯−1 and 1−y¯; y¯−1≤Conc.I≤1−y¯ when y is dichotomous [41]. Conc. I acquires a negative value when the curve lies above the line of equality, which indicates a disproportionately lower prevalence of CC screening knowledge and utilisation of services among the poor (i.e., pro-poor). A positive value of Conc. I signifies a higher concentration of health indicators among the rich (i.e., pro-rich). There is no socioeconomic inequality in the distribution of CC screening knowledge and utilisation of services (y) when the value of Conc. I is zero and the concentration curve coincides with the 45° line. The dichotomous character of the knowledge and utilisation of CC screening services may result in unstable bounds in response to varying means; therefore, the normalised standard index was estimated to check the robustness of the estimation [42, 43]. In addition, when the outcome variable is dichotomous, the Conc. I has to be corrected in order to allow comparisons between groups of individuals from different time periods that may show different levels of use of health services [45]. In the context of a dichotomous outcome variable, the Erreygers’s Conc. I is the Conc. I multiplied by four times the mean health or outcome of interest [45]. Erreygers’ suggested corrected CI can be expressed as: where ymax and ymin are the boundary of y (knowledge and utilisation of CC screening services). When the Erreygers’ corrected index is used, the decomposition of inequality is generally expressed as: This estimate produces an index that satisfies various attractive axiomatic properties for an inequality index, including the sign condition, scale invariance and mirror properties [46, 47]. The adjusted Conc. I method allows for an examination of the causes of (and their corresponding contributions to) and levels of changes in inequalities in terms of knowledge and utilisation of CC screening services [40]. In addition, multiple logistic regression was applied to measure the likelihood of CC screening knowledge, awareness and utilisation of services. Adjusted odds ratios (AORs) with a 95% confidence interval (CI) were estimated for identifying influencing factors on CC screening knowledge and utilisation of services at a 5% or lower level of significance. All the estimates were considered by sampling weights according to the DHS guideline. According to the DHS guideline, sample weights are estimated to six decimals but are presented in the standard recode files without the decimal point. They need to be divided by 1,000,000 before use to approximate the number of cases. As part of complex sample parameters when standard errors, confidence intervals or significance testing is required for the indicator [54]. All statistical analyses were performed with Stata/SE-13 software (StataCorp, College Station, TX, USA).
N/A