Sage Journals: Discover world-class research

Abstract

Background:

Access to family planning (FP) information is vital for reproductive autonomy and the Sustainable Development Goals. In Somaliland, modern contraceptive uptake remains critically low. While sociodemographic barriers are known, geospatial dimensions of information accessibility remain unexplored. This study investigates individual, community, and spatial determinants of women’s exposure to FP messages.

Methods:

We analyzed data from the 2020 Somaliland Health and Demographic Survey among women aged 15–49. We employed multilevel mixed-effects logistic regression to assess predictors of media exposure. Additionally, spatial analysis techniques including Global Moran’s I and Getis-Ord Gi* statistics were utilized to identify geographic clusters (“hotspots” and “cold spots”) of exposure.

Results:

Only 25.28% of women reported exposure to FP messages, primarily via TV and radio; mobile dissemination was underutilized (11%). Multilevel analysis indicated that higher education (adjusted odds ratios [AOR] = 6.48), wealth, and internet usage significantly increased exposure odds. Conversely, nomadic populations (AOR = 0.38) and rural residents faced severe exclusion. Spatial analysis revealed a sharp East–West divide: significant “hotspots” clustered in western regions (Awdal, Marodijeh), while significant “cold spots” were identified in the eastern region of Sool.

Conclusions:

A profound “digital and developmental divide” characterizes reproductive health communication in Somaliland. Current strategies benefit urban, educated women in the west, leaving nomadic populations in an information vacuum. Policy must shift to spatially targeted interventions, utilizing community outreach in identified “cold spots” and voice-based mobile technology to reach uneducated and nomadic women.

Keywords

family planning mass media exposure spatial analysis Somaliland health geographics women’s health promotion

Introduction

Family planning (FP) is a fundamental human right and a pivotal public health intervention that empowers individuals and couples to decide freely the number and spacing of their children, thereby directly influencing fertility regulation and reproductive well-being. Conceptually, the utilization of FP methods is intrinsically aligned with the United Nations Sustainable Development Goals (SDGs), specifically contributing to the eradication of poverty (SDG 1), the promotion of good health by reducing maternal and infant mortality (SDG 3), and the fostering of gender equality and women’s empowerment (SDG 5).^1–3 Access to modern contraceptives enables women to pursue educational and economic opportunities, which is essential for achieving the demographic dividend in developing nations.^4–6 Furthermore, the dissemination of FP messages through mass media is recognized as a vital strategy to bridge the knowledge gap, build self-efficacy, and promote positive behavioral changes regarding reproductive health.^7–9 Despite these established benefits, a significant unmet need for contraception persists in Sub-Saharan Africa, where structural and sociocultural barriers often hinder access to information and services, necessitating a rights-based approach to care.^10–12

Globally, the history of FP uptake reveals stark disparities between developed and developing regions, with Sub-Saharan Africa bearing the highest burden of maternal mortality and fertility rates. Historical data indicate that while global contraceptive prevalence has increased, regions such as West and Central Africa continue to lag, often due to inadequate exposure to health communication.^13–15 For instance, in countries like the Democratic Republic of Congo and Nigeria, high maternal mortality rates are exacerbated by low contraceptive usage and limited access to mass media messages, particularly in rural areas.^10,16,17 Recent studies across 26 Sub-Saharan African countries indicate that exposure to FP messages significantly increases contraceptive use, yet prevalence varies widely, with countries like Chad showing exposure rates as low as 11%.^18–20 In East Africa, while countries like Kenya have seen improvements, significant gaps remain in ensuring universal access to reproductive health information, a challenge mirrored in Somaliland’s context, where traditional methods often prevail over modern contraceptives due to a lack of awareness programs.^17,21–23

Previous studies have extensively documented the sociodemographic factors and media exposure variables associated with the uptake of FP methods. Research consistently demonstrates that women exposed to FP messages via mass media such as radio, television, and mobile phones are significantly more likely to utilize modern contraception compared with those with no exposure.^6,7,22,24,25 However, this relationship is mediated by various factors; for example, younger women, those with lower educational attainment, and those residing in rural areas often face structural barriers to accessing these messages.^4,9,26 In Ghana and Sierra Leone, studies have shown that men’s exposure to FP messages also plays a crucial role in consistent condom use and couple communication, highlighting the importance of male involvement in reproductive decisions.^6,14,27 Furthermore, socioeconomic status is a critical determinant; wealthier women and those with higher decision-making autonomy are more likely to act on media messages and adopt modern methods.^11,15,28,29 Conversely, misconceptions and negative attitudes toward contraception, often fueled by a lack of accurate information, continue to hinder uptake in many communities.^10,28,30

To understand these issues better, this study uses a spatial and multilevel analysis of data from the Demographic and Health Survey (DHS). This method allows us to look at data from both urban and rural areas to understand what drives exposure to FP messages in Somaliland, similar to methods used in other major studies.^19,21,31 By using a multilevel model, we can separate individual factors and community factors.^21,22,32 Despite previous research, there is still a gap in understanding the specific location-based factors that affect media exposure in Somaliland. This study contributes new knowledge by focusing specifically on “exposure to media” as a key tool to solve high fertility and maternal death rates. The significance of this research is that it provides clear evidence to help leaders create better, targeted awareness campaigns to improve health outcomes.^7,11,14,17

Materials and Methods

Study area and data source

The study was conducted in the Republic of Somaliland, covering the six major administrative regions: Awdal, Marodijeh, Sahil, Togdheer, Sool, and Sanaag. These regions encompass a diverse range of settlement patterns, including urban centers, rural villages, and nomadic settlements, characterized by varying degrees of health care infrastructure and information accessibility. The data for this analysis were derived from the Somaliland Health and Demographic Survey (SLHDS) 2020, the first nationally representative health survey conducted in the region.³³ The SLHDS employed a cross-sectional study design with a stratified multi-stage cluster sampling technique. In the first stage, Primary Sampling Units defined as enumeration areas (EAs) were selected with probability proportional to size. In the second stage, households were selected using systematic random sampling from the listed households within each EA. For nomadic populations, a specific strategy involving temporary nomadic settlements was utilized to ensure inclusivity. The final analytic sample consisted of women of reproductive age (15–49 years) after data cleaning and variable restructuring were performed to remove missing values and inconsistencies.^33,34

Study variables

The outcome variable of the study was exposure to FP messages. This binary variable was derived from survey questions asking respondents if they had heard or seen messages about FP in the months preceding the survey through mass media channels, including radio, television, newspapers, social media, or mobile phones. The variable was coded as “1” if the respondent had been exposed to FP messages via any of these sources and “0” otherwise.³³ For the purposes of this study, “FP” encompasses all modern contraceptive methods, including male condoms. The independent variables were categorized into two levels: individual and community levels. Individual factors included the age of the respondent, educational attainment, marital status, household wealth index, and media ownership (radio, TV, and mobile phone). Community-level variables included the region, place of residence (urban, rural, nomadic), and wealth.³⁴

Statistical analysis

Data analysis was conducted using STATA and R Studio. Initially, descriptive statistics were used to summarize the demographic characteristics of the sample. Bivariate analysis was performed to assess the association between independent variables and FP message exposure using the Chi-square to account for the complex survey design. Variables with a p-value < 0.05 were considered statistically significant and included in the subsequent analysis. To account for the hierarchical nature of the data, where women are nested within clusters, a multilevel mixed-effects logistic regression model was fitted. Four models were estimated: the Null Model (Model 0) containing only the outcome variable to assess cluster variation; Model I containing individual-level factors; Model II containing community-level factors; and Model III (the full model) adjusting for both levels simultaneously. Model fitness was compared using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Log-Likelihood (LL), while the Intraclass Correlation Coefficient (ICC) was calculated to quantify the variance attributable to community differences.^34,35

Spatial analysis

Spatial analysis was conducted using R Studio (utilizing packages such as ggplot2 and tmap) to evaluate the geographic distribution of FP message exposure across the SLHDS clusters. The units of analysis for the spatial maps represent the predicted probability of exposure to FP messages derived from the smoothed prevalence rates across clusters. The Global Moran’s I statistic was employed to test for spatial autocorrelation, determining whether the exposure patterns were clustered, dispersed, or random across the study area.³⁶ A statistically significant Moran’s I (p < 0.05) indicated the presence of spatial clustering. To identify specific local clusters, Local Moran’s I (LISA) and Getis-Ord Gi* statistics were used. These analyses pinpointed “hotspots” (areas with significantly high exposure) and “cold spots” (areas with significantly low exposure) of FP messaging, providing visual evidence of regional disparities.^34,37

Results

Proportion of exposure FP messages

As shown in Table 1, the distribution of exposure status was heavily skewed. The prevalence of exposure to FP messages was low, with only 25.28% ( $S E = 1 %$ ) of participants reporting contact with such media. The remaining 74.72% (95% CI [72.78, 76.56]) were classified as unexposed.

Table 1.

Proportion of Exposure Planning Messages

Exposure status	Proportion	Percentage	Standard error	(95% conf. Interval)
Unexposed	0.7472	74.72%	0.0097	0.7278–0.7656
Exposed	0.2528	25.28%	0.0097	0.2344–0.2722

Prevalence of FP messages channels

A disaggregated analysis of media sources indicates a distinct hierarchy in the dissemination of FP information, favoring traditional broadcast media over digital or print formats. As illustrated in Figure 1, audio-visual channels specifically television (26%) and radio (19%) demonstrated significantly higher efficacy in reaching the target demographic compared with text-based mediums. Despite the global proliferation of digital technology, social media (9%) and mobile phones (11%) remain underutilized as conduits for health communication in this context. The negligible reach of Newspapers/Magazines (3%) suggests that “print media is currently an ineffective vehicle for mass sensitization regarding FP.”

FIG. 1.

Prevalence of family planning message exposure by specific media channels among women in Somaliland.

Univariate and bivariate analysis of exposure of FP messages among women in Somaliland

A total of 1979.3 (weighted count) women of reproductive age were included in the analysis. The distribution of participants generally reflects the population density of Somaliland, with the majority of respondents residing in the Marodijeh (33.0%) and Togdheer (26.1%) regions, which house the major urban centers. Fewer participants were drawn from Sahil (6.6%), Awdal (10.0%), and Sool (9.2%), corresponding to the lower population densities in these areas. The descriptive statistics profile the study population as predominantly young and urban-centric. The modal age cohort is 25–29 years (25.5%). Socioeconomic indicators reveal profound structural vulnerabilities: educational attainment is negligible, with 77.34% of participants lacking formal schooling. Furthermore, while 51.7% of partners were employed, female labor force participation is virtually nonexistent, with 98.8% of respondents reporting unemployment. Wealth distribution is polarized, with significant concentrations in the highest (35.0%) and lowest (24.3%) quintiles. Technological access presents a paradox between hardware ownership and connectivity; while mobile phone ownership is widespread (77.5%), internet penetration remains low (12.7%). Engagement with traditional mass media is similarly limited, as the vast majority report zero consumption of radio (91.5%) or television (73.3%). These structural factors coexist with deep-seated conservatism, evidenced by a critically low contraceptive prevalence rate (3.0%) and prevalent justification of intimate partner violence among roughly 40%–47% of respondents.

Bivariate analysis demonstrated that exposure to FP messages is significantly shaped by media access, socioeconomic standing, and geography, rather than by individual employment or attitudes. The strongest predictors were media consumption habits, particularly television viewing, which yielded the highest association ( $χ^{2} = 595.81, p < 0.001$ ); weekly viewers had an 85.6% exposure rate compared with 22.7% for nonviewers. Similar strong associations were observed for radio listening and internet usage. Socioeconomic factors showed a clear positive gradient; respondents in the highest wealth quintile ( $χ^{2} = 430.76, p < 0.001$ ) and those with higher education were far more likely to be exposed than those with lower status. Geographic disparities were also pronounced, with urban residents (57.5%) showing significantly higher exposure than nomadic populations (7.8%), $χ^{2} = 379.39, p < 0.001$ . In contrast, the respondent’s own employment status ( $p = 0.1694$ ), contraceptive use ( $p = 0.085$ ), and attitudes toward wife-beating ( $p > 0.05$ ) showed no statistically significant relationship with exposure (Table 2).

Table 2.

Univariate and Bivariate Analysis of Exposure of Family Planning Messages among Women in Somaliland

Variable	Frequency (n)	Exposure family planning messages		DF	Chi-square (χ²)	p value
Variable	Frequency (n)	Unexposed (%)	Exposed (%)	DF	Chi-square (χ²)	p value
Age Group				6	39.88	0.0325
15–19	113 (6.3%)	81.7	18.3
20–24	285.3 (15.9%)	64.2	35.8
25–29	458.5 (25.5%)	55	45
30–34	356.1 (19.8%)	57.9	42.1
35–39	285.4 (15.89%)	67.6	32.4
40–44	182.1 (10.1%)	61.8	38.2
45–49	115.9 (6.5%)	58.7	41.3
Education Level				3	182.6	<0.001
No Education	1389.3 (77.34%)	69	31
Primary	296.1 (16.5%)	42.1	57.9
Secondary	66.8 (3.7%)	30.5	69.5
Higher	44.1 (2.5%)	7.5	92.5
Husband/Partner Worked				1	242.12	<0.001
No	867.9 (48.3%)	79	21
Yes	928.4 (51.7%)	45.4	54.6
Respondent Worked				1	7.82	0.1694
No	1775 (98.8%)	62	38
Yes	21.2 (1.2%)	34	66
Contraceptive Use				1	7.39	0.085
User	54.5 (3.0%)	45	55
Non-user	1741.8 (97.0%)	62.2	37.8
Ever Used internet				1	155.18	<0.001
No	1567.2 (87.3%)	66.8	33.2
Yes	229.1 (12.7%)	26.4	73.6
Owns Mobile Phone				1	67.78	<0.001
No	404 (22.5%)	78.2	21.8
Yes	1392.3 (77.5%)	56.8	43.2
Wife Beating (Refuses Sex)				1	6.98	0.1242
Not Justified (No)	951.3 (53.0%)	64.3	35.7
Justified (Yes)	845 (47.0%)	58.6	41.4
Wife Beating (Argues)				1	0.01	0.9581
Not Justified (No)	1075.4 (59.9%)	61.6	38.4
Justified (Yes)	720.9 (40.0%)	61.8	38.2
Frequency Listening to Radio				2	95.26	<0.001
At least once a week	114 (6.4%)	22.3	77.7
Less than once a week	38.4 (2.1%)	47	53
Not at all	1643.9 (91.5%)	64.7	35.3
Frequency Watching TV				2	595.81	<0.001
At least once a week	390.1 (21.7%)	14.4	85.6
Less than once a week	90 (5.0%)	37	63
Not at all	1316.1 (73.3%)	77.3	22.7
Wealth Quintile				4	430.76	<0.001
Lowest	436.8 (24.3%)	87.6	12.4
Second	202.7 (11.3%)	84.2	15.8
Middle	190 (10.6%)	71.2	28.8
Fourth	339.9 (19.0%)	60.9	39.1
Highest	626.8 (35.0%)	33.7	66.3
Residence				2	379.39	<0.001
Urban	946.7 (52.7%)	42.5	57.5
Rural	453.4 (25.2%)	74.8	25.2
Nomadic	396.2 (22.1%)	92.2	7.8
Region				5	352.91	<0.001
Awdal	178.5 (10.0%)	68	32
Marodijeh	592.8 (33.0%)	34.7	65.3
Sahil	118.7 (6.6%)	56.9	43.1
Togdheer	469.3 (26.1%)	72.7	27.3
Sool	165.3 (9.2%)	88.7	11.3
Sanaag	271.8 (15.1%)	82.8	17.2

Model comparison

In the multilevel analysis, the null model (Model 0) revealed substantial clustering, with 47.2% of the variance in exposure to FP messages attributed to community-level factors (ICC = 0.472; variance = 2.94). This strong clustering was further corroborated by a high median odds ratio (MOR) of 5.13, indicating significant heterogeneity between clusters. The inclusion of explanatory variables in subsequent models progressively reduced this unobserved heterogeneity, with the final model (Model 3) effectively explaining the between-cluster variance (ICC ≈ 0.0%; MOR = 1.0). Consequently, Model 3 was selected as the best-fitting model, exhibiting the lowest AIC = 1673.7 and BIC = 1853.3, alongside the highest LL (−804.86) (Table 3).

Table 3.

Model Comparison

Random effects and model fit	Model 0	Model 1 (AOR [95% CI])	Model 2 (AOR [95% CI])	Model 3 (AOR [95% CI])
Measure of variance
Variance (var(_cons))	2.94	0.49	0.03	∼0.00
ICC (Intraclass Correlation)	0.472	0.13	0.01	∼0.0%
MOR	5.13	1.95	1.18	1.00
Model fit statistics
Log Likelihood	−1026.61	−861.51	−884.22	−804.86
AIC	2057.2	1767	1794.4	1673.7
BIC	2068.4	1890.5	1867.4	1853.3

AIC, Information Criterion; AOR, adjusted odds ratio; BIC, Bayesian Information Criterion; CI, confidence interval; ICC, Intra-Cluster Correlation.

Multilevel regression analysis

The multilevel analysis examined various factors associated with exposure to FP messages, with adjusted odds ratios (AOR) and 95% confidence intervals (CI) presented across multiple models as shown in Table 4. Regional variations were evident. Marodijeh, which houses the capital city Hargeisa and represents the most urbanized and wealthy region demonstrated significantly higher odds of exposure (AOR = 1.86, 95% CI [1.21, 2.86]) compared with the reference category, Awdal. Similarly, Sahil showed high odds (AOR = 1.95). Conversely, the more remote and pastoralist regions of Sool (AOR = 0.47) and Sanaag (AOR = 0.51) had significantly lower odds of receiving these messages. Residence type also played a critical role, with nomadic populations (AOR = 0.38, 95% CI [0.22, 0.66]) and rural residents (AOR = 0.55, 95% CI [0.40, 0.76]) exhibiting decreased odds of exposure relative to urban residents. Age showed a consistent trend, with older women, particularly those aged 45–49 years (AOR = 4.39, 95% CI [2.05, 9.39]), exhibiting significantly greater odds of encountering FP messages relative to those aged 15–19 years. Education level also had a substantial impact, as women with higher education (AOR = 6.48, 95% CI [1.70, 24.68]) and primary education (AOR = 1.78, 95% CI [1.25, 2.55]) had significantly increased odds of exposure compared with those with no education. Employment status was another relevant factor, with respondents whose husbands worked (AOR = 1.71, 95% CI [1.28, 2.28]) demonstrating significantly higher odds of accessing FP information relative to those whose partners did not work. Media consumption habits were strongly associated with the likelihood of exposure; internet users (AOR = 1.72, 95% CI [1.05, 2.81]) had higher odds, while lack of traditional media access was associated with reduced odds, specifically for those who did not listen to the radio (AOR = 0.29, 95% CI [0.16, 0.52]) or watch TV (AOR = 0.26, 95% CI [0.17, 0.40]) at all compared with weekly consumers. Overall, the findings highlight significant regional, sociodemographic, and media-related factors influencing exposure to FP messages.

Table 4.

Multivariate Analysis of Exposure Family Planning Messages and Associated Factors

Variable	Model 1 (AOR [95% CI])	Model 2 (AOR [95% CI])	Model 3 (AOR [95% CI])
Individual Level Factors
Age Group
15–19	1		1
20–24	2.34 [1.22, 4.51]*		2.33 [1.22, 4.46]**
25–29	3.12 [1.67, 5.84]***		2.80 [1.50, 5.22]**
30–34	2.73 [1.44, 5.19]**		2.55 [1.35, 4.83]**
35–39	2.50 [1.29, 4.85]**		2.14 [1.11, 4.14]*
40–44	3.41 [1.69, 6.88]**		2.93 [1.46, 5.89]**
45–49	5.00 [2.34, 10.69]***		4.39 [2.05, 9.39]***
Education
No Education	1		1
Primary	1.73 [1.21, 2.46]**		1.78 [1.25, 2.55]**
Secondary	1.81 [0.86, 3.83]		1.65 [0.79, 3.43]
Higher	6.72 [1.79, 25.28]**		6.48 [1.70, 24.68]**
Husband Worked
No	1		1
Yes	2.58 [1.97, 3.39]***		1.71 [1.28, 2.28]***
Respondent Worked
No	1		1
Yes	5.45 [1.45, 20.44]*		2.95 [0.76, 11.43]
Internet Use
No	1		1
Yes	1.94 [1.17, 3.22]*		1.72 [1.05, 2.81]*
Own Mobile
No	1		1
Yes	1.21 [0.88, 1.66]		1.20 [0.87, 1.66]
Radio Frequency
At least once a week	1		1
Less than once a week	0.22 [0.07, 0.66]**		0.24 [0.08, 0.68]**
Not at all	0.25 [0.14, 0.45]***		0.29 [0.16, 0.52]***
TV Frequency
At least once a week	1		1
Less than once a week	0.45 [0.22, 0.92]*		0.61 [0.29, 1.26]
Not at all	0.13 [0.09, 0.19]***		0.26 [0.17, 0.40]***
Contraceptive Use
User	1		1
Non-user	0.40 [0.20, 0.80]**		0.56 [0.28, 1.11]
Attitude: Wife Beating (Argues)
Justified (No)	1		1
Justified (Yes)	1.16 [0.81, 1.65]		1.09 [0.76, 1.57]
Attitude: Wife Beating (Refuses Sex)
Justified (No)	1		1
Justified (Yes)	1.26 [0.88, 1.79]		1.31 [0.92, 1.88]
Community Level Factors
Wealth Quintile
Lowest		1	1
Second		0.79 [0.47, 1.33]	0.76 [0.45, 1.28]
Middle		1.59 [0.92, 2.76]	1.35 [0.77, 2.35]
Fourth		1.50 [0.87, 2.57]	1.11 [0.64, 1.92]
Highest		4.36 [2.53, 7.52]***	1.68 [0.97, 2.91]
Residence
Urban		1	1
Rural		0.42 [0.31, 0.58]***	0.55 [0.40, 0.76]***
Nomadic		0.22 [0.13, 0.38]***	0.38 [0.22, 0.66]***
Region
Awdal		1	1
Marodijeh		2.10 [1.36, 3.24]***	1.86 [1.21, 2.86]**
Sahil		1.76 [1.15, 2.68]**	1.95 [1.28, 2.98]**
Togdheer		0.49 [0.31, 0.75]**	0.54 [0.34, 0.85]**
Sool		0.40 [0.25, 0.65]***	0.47 [0.28, 0.79]**
Sanaag		0.44 [0.27, 0.72]**	0.51 [0.33, 0.79]**

AOR, adjusted odds ratio, *p value < 0.05, **p value < 0.01, ***p value < 0.001.

Spatial distribution of exposure to FP messages

In the context of Somaliland, spatial analysis is instrumental in uncovering the geographical distribution and variations in exposure to FP messages across its regions, as depicted in the map. This analytical approach helps pinpoint areas with significant information gaps, thereby informing focused public health interventions. As illustrated in Figure 2, the spatial distribution reveals notable regional disparities. Consistent with the regression results, the Marodijeh region, being the urban center, and the Sahil region exhibit the highest relative likelihood of exposure. Conversely, regions such as Sool show the lowest reported levels (0.113–0.170), represented by the lightest shading, with Sanaag also demonstrating limited reach (0.170–0.180). This geographical insight is vital for tailoring communication campaigns and resource allocation to ensure that FP information effectively reaches underserved populations in the eastern regions.

FIG. 2.

Spatial distribution map illustrating the predicted probability of exposure to family planning messages across Somaliland regions.

Global Moran’s I spatial autocorrelation

A Global Moran’s I test was performed to assess the spatial autocorrelation in the distribution of exposure to FP messages across the regions. As presented in the accompanying Table 5 and Figure 3, the test yielded a Moran’s Index of 0.26088, a Z-score of 1.87512, and a p-value of 0.030388. These statistics indicate a positive spatial autocorrelation, suggesting that regions with similar levels of message exposure tend to cluster together geographically. With a p-value below the standard significance threshold (p < 0.05), the spatial autocorrelation is statistically significant, implying that the observed clustering is unlikely to be a result of random chance. This finding underscores that the distribution of exposure to FP messages is not random and points to underlying spatial dependencies that warrant further investigation into the factors contributing to these regional clusters.

FIG. 3.

Global Moran’s I scatter plot showing spatial autocorrelation of media exposure.

Table 5.

Global Moran’s I

Statistic	Moran’s I statistic	Z-Score	p value
Value	0.26088	1.87512	0.030388

Local Moran’s I spatial autocorrelation

In Figure 4, the Local Moran’s I analysis reveals distinct spatial patterns in exposure to FP messages, though statistical significance varies by region. While Maroodi Jeex exhibited the strongest autocorrelation coefficient (0.314–0.930), suggesting potential clustering, this pattern was not statistically significant ( $p \geq 0.05$ ). Conversely, Awdal and Sool demonstrated both positive spatial autocorrelation (0.188–0.314) and statistical significance ( $p < 0.05$ ), confirming them as distinct, nonrandom spatial clusters likely representing a “hotspot” of high exposure in the west and a “cold spot” in the east, respectively. In contrast, Togdheer, Sahil, and Sanaag displayed low autocorrelation values and nonsignificance, indicating that exposure distributions in these central and coastal areas are random and driven by localized factors rather than broader regional dependence.

FIG. 4.

Local Moran’s I (LISA) cluster map identifying significant spatial clusters (hotspots and coldspots) of exposure.

Hotspots and coldspots analysis (Getis-Ord Gi*)

The Getis-Ord Gi* statistic was employed to identify statistically significant spatial clusters of high values (hot spots) and low values (cold spots) regarding exposure to FP messages. The analysis in Figure 5 reveals a sharp geographical polarization. A distinct “hot spot” was identified in the western regions of Awdal and Maroodi Jeex, indicated by the highest Gi* values (1.397–1.969) and dark green shading. This signifies that these areas not only have high exposure rates but are also surrounded by neighbors with similarly high exposure. Conversely, the eastern region of Sool emerged as a significant “cold spot” (red shading), exhibiting the lowest Gi* statistics (−1.786 to −0.946). This indicates a spatial cluster where low exposure levels are concentrated. Sanaag also displayed negative values, suggesting a tendency toward a cold spot, while Togdheer (yellow) showed near-zero values (−0.031 to 1.102), indicating a random spatial distribution with no significant clustering. These findings statistically confirm the East–West divide, highlighting where intervention is most critically needed.

FIG. 5.

Getis-Ord Gi* hotspot analysis identifying statistically significant geographic polarization of family planning information access.

Discussion

This study provides the first comprehensive spatial and multilevel analysis of exposure to FP messages among women in Somaliland, utilizing data from the 2020 SLHDS. The findings reveal a critical information gap, with only 25.28% of women of reproductive age reporting exposure to FP messages through mass media. This low prevalence underscores a significant public health challenge, as lack of information is a primary barrier to contraceptive uptake and reproductive autonomy.^10,12 This finding aligns with broader trends in Sub-Saharan Africa, where exposure rates vary significantly but often remain inadequate in regions with high maternal mortality.^19,21,38

The analysis of media channels highlights a distinct hierarchy favoring traditional broadcast media. Television and radio were the primary sources of information, while print media reach was negligible (3%), likely due to high illiteracy rates (77.34% of respondents had no education). Paradoxically, while mobile phone ownership was widespread (77.51%), exposure to FP messages via mobile phones was low (11%). This represents a significant “missed opportunity” for mHealth interventions. Unlike studies in other developing nations where mobile technology has successfully bridged health information gaps,^6,11 Somaliland has yet to fully leverage its high mobile penetration for reproductive health advocacy.

While previous literature generally supports a strong link between media exposure and contraceptive uptake, our bivariate analysis showed a non-significant association (p = 0.085) between exposure and usage in this population. This suggests that in the context of Somaliland, while media effectively generates awareness, the translation of this awareness into actual usage is likely hindered by other structural barriers such as limited access to services and deep-seated cultural norms rather than a lack of information alone. Therefore, media exposure should be viewed as a necessary prerequisite for behavior change, but not a standalone solution.

Sociodemographic factors played a decisive role in determining exposure. Consistent with previous research in Ethiopia, Nigeria, and Uganda, women with higher educational attainment and those in the highest wealth quintiles were significantly more likely to encounter FP messages.^4,9,15 This suggests that current communication strategies are inadvertently favoring the socioeconomic elite, leaving the most vulnerable populations, the poor and uneducated behind. Furthermore, the study found that older women (45–49 years) had higher odds of exposure compared with adolescents (15–19 years). This is concerning given that adolescents often face the highest risks regarding reproductive health and have a critical need for information to delay first pregnancies.^4,6

The multilevel analysis revealed that community-level factors accounted for a substantial portion (47.2%) of the variance in message exposure, a finding consistent with studies emphasizing the influence of neighborhood context on health behaviors.^21,32 The most striking disparities were geographic. Women in urban areas were far more likely to be exposed than those in rural or nomadic settlements. This disparity is clearly reflected in the regional results, where Marodijeh which houses the capital city Hargeisa and represents the nation’s economic and administrative hub showed significantly higher exposure odds compared with the peripheral and pastoralist-dominated regions of Sool and Sanaag. Nomadic women, who constitute a significant portion of the population, showed the lowest odds of exposure (AOR = 0.38). This aligns with the “structural barriers” hypothesis, where remote populations are systematically excluded from information flows due to infrastructure deficits.^26,34

Spatial analysis further pinpointed these disparities, identifying a clear East–West divide. “Hotspots” of high exposure were clustered in the western regions of Awdal and Marodijeh, while significant “cold spots” were identified in the eastern region of Sool. This spatial clustering, confirmed by the Global Moran’s I and Getis-Ord Gi* statistics, suggests that information dissemination is not random but is heavily influenced by regional development and administrative reach.^34,36,37 The low exposure in Sool and Sanaag may be attributed to lower media infrastructure density and potential accessibility challenges compared with the more urbanized western corridor.

Finally, the study noted that women whose husbands were employed had higher odds of exposure. This reinforces the literature suggesting that male involvement and household economic stability facilitate access to information channels.^16,27 However, individual employment and attitudes toward domestic violence were not significant predictors in the multivariate model, suggesting that structural access to media (wealth, location, device ownership) outweighs individual attitudinal factors in determining whether a woman receives health messages.

Strength and limitation

The primary strength of this study lies in its use of the SLHDS 2020, the first nationally representative health survey in Somaliland, which ensures the high generalizability of the findings to the broader population. Methodologically, the integration of multilevel mixed-effects modeling rigorously accounts for hierarchical data clustering, correcting for intra-class correlation. Furthermore, the application of spatial statistics specifically Global Moran’s I and Getis-Ord Gi* uniquely moves beyond national averages to visualize sub-national disparities, effectively identifying precise geographic “cold spots” to guide targeted public health interventions.

However, the study is subject to limitations inherent to its cross-sectional design, which allows for the determination of associations but not causality. The reliance on self-reported data introduces potential recall bias, as respondents may not remember incidental exposure to messages. Additionally, the binary outcome variable captures only the presence of exposure, not the frequency or quality. Furthermore, this study focuses specifically on mass media exposure and does not account for FP messages received through interpersonal channels, such as community health workers, religious leaders, or peer-to-peer discussions, which are also significant sources of information in this context. Finally, the absence of geospatial data on media infrastructure, such as signal coverage maps, limits the ability to definitively explain the structural causes of the observed regional disparities.

Conclusion

This study highlights a profound inequity in the dissemination of FP information in Somaliland. With three-quarters of women unexposed to FP messages, the current communication landscape is insufficient to drive the behavioral changes necessary to achieve Sustainable Development Goals 3 and 5. The analysis confirms that exposure is driven by a “digital and developmental divide,” where wealthy, educated, urban women in the western regions are the primary recipients of health information, while poor, uneducated, nomadic women in the eastern regions are left in an information vacuum.

Policy Implications

To bridge this gap, the Ministry of Health Development and international partners must shift from a “one-size-fits-all” approach to targeted spatial interventions. 1

Geographic Targeting: Immediate priority should be given to the “cold spots” identified in Sool and Sanaag, utilizing community-based outreach rather than relying solely on mass media, which does not reach these areas effectively.

Leveraging Mobile Technology: Given the high rate of mobile phone ownership but low message exposure, there is an urgent need to implement SMS and Interactive Voice Response campaigns. These should be voice-based to bypass literacy barriers affecting the 77% of women with no formal education.

Tailored Content for Nomads: Specific strategies must be designed for nomadic populations, such as utilizing portable radio distribution or integrating FP education into veterinary and water point services.

Targeting Men and Community Leaders: Because household dynamics significantly influence information access (as indicated by the significance of husband’s employment), communication strategies must explicitly target men to create an enabling environment. Messages should be framed to appeal to male responsibilities regarding family health and economic stability, thereby reducing domestic barriers to radio/TV access for women.

Localization through Religious Networks: To penetrate the “Cold Spots” in the eastern regions effectively, FP messages should be integrated into Friday sermons and disseminated by local religious leaders. Framing birth spacing within the context of Islamic teachings (e.g., breastfeeding for 2 years) can legitimize the messages for conservative populations who might ignore standard mass media.

By addressing these spatial and structural disparities, Somaliland can enhance reproductive health literacy, empower women to make informed choices, and ultimately improve maternal and child health outcomes.

Ethics Approval and Consent to Participate

Ethical board review and participant consent were not specifically required for the present analysis due to its nature as a secondary analysis of publicly accessible data from the SHDS. Access to the requisite data was formally requested by the authors and subsequently approved by the SHDS program administrators.

Ethical Disclosures

Authors got permission from Health and Demographic Surveys (DHS) Program and downloaded data from this link (https://microdata.nbs.gov.so/index.php/catalog/50). As this data is publicly available and has no personal identifiers, Ethical approval was not necessary.

Consent for Publication

Not applicable.

Authors’ Contributions

M.M.A. conceived and designed the study, performed the data extraction, methodology, and data analysis, and drafted the initial article. K.J.O. contributed to the data analysis and the writing of the discussion and conclusion sections. A.A.M. was involved in data analysis and article editing. Y.D.A. contributed to writing the discussion. I.F.M. contributed to the methodology section. A.H.M. supervised the research, reviewed, and edited the final article. All authors read and approved the final version of the article for publication.

Footnotes

Availability of Data and Materials

The dataset utilized in this research can be obtained by contacting the corresponding author with a legitimate request.

Author Disclosure Statement

The authors declare that they have no competing interests.

Funding Information

The authors received no specific funding for this work.

Abbreviations

References

Ali

, Ali

, Begum

. Utilization of family planning methods among currently married women in Pakistan: Insight from Pakistan Demographic and Health Survey 2017–2018. Health Care Women Int, 2024; 45(1):129–143.

Pradhan

, Costa

, Rybski

, et al. A systematic study of sustainable development goal (SDG) interactions. Earth’s Futur, 2017; 5(11):1169–1179.

Starbird

, Norton

, Marcus

. Investing in family planning: Key to achieving the sustainable development goals. Glob Health Sci Pract, 2016; 4(2):191–210.

Ahmed

, Seid

. Association between exposure to mass media family planning messages and utilization of modern contraceptive among urban and rural youth women in Ethiopia. Int J Womens Health, 2020; 12:719–729.

Canning

, Schultz

. The economic consequences of reproductive health and family planning. Lancet, 2012; 380(9837):165–171.

Sserwanja

, Turimumahoro

, Nuwabaine

, et al. Association between exposure to family planning messages on different mass media channels and the utilization of modern contraceptives among young women in Sierra Leone: Insights from the 2019 Sierra Leone Demographic Health Survey. BMC Womens Health, 2022; 22(1):376.

Abita

, Girma

. Exposure to mass media family planning messages and associated factors among youth men in Ethiopia. Heliyon, 2022; 8(9):e10544.

Babalola

, Figueroa

, Krenn

. Association of mass media communication with contraceptive use in Sub-Saharan Africa: A meta-analysis of Demographic and Health Surveys. J Health Commun, 2017; 22(11):885–895.

Birabwa

, Chemonges

, Tetui

, et al. Knowledge and information exposure about family planning among women of reproductive age in informal settlements of Kira Municipality, Wakiso District, Uganda. Front Glob Women’s Heal, 2021; 2:650538.

10.

Anfaara

, Antabe

, Sano

. Postpartum Married Women’s Mass Media Exposure to Family Planning Messages in the Democratic Republic of Congo: The Role of Women’s Household Decision-Making Autonomy. In: Health Geography in Sub-Saharan Africa: Development-Health Nexus. Springer; 2023:85–98.

11.

Ghose

, Adjei

, Yaya

. Exposure to family planning messages on social media and its association with maternal healthcare services in Mauritania. BMC Womens Health, 2024; 24(1):533.

12.

Hardee

, Kumar

, Newman

, et al. Voluntary, human rights–based family planning: A conceptual framework. Stud Fam Plann, 2014; 45(1):1–18.

13.

Cleland

, Conde-Agudelo

, Peterson

, et al. Contraception and health. Lancet, 2012; 380(9837):149–156.

14.

Das

, Samad

, Al Banna

, et al. Association between media exposure and family planning in Myanmar and Philippines: Evidence from nationally representative survey data. Contracept Reprod Med, 2021; 6(1):11.

15.

Konkor

, Sano

, Antabe

, et al. Exposure to mass media family planning messages among post-delivery women in Nigeria: Testing the structural influence model of health communication. Eur J Contracept Reprod Health Care, 2019; 24(1):18–23.

16.

Hogan

, Foreman

, Naghavi

, et al. Maternal mortality for 181 countries, 1980–2008: A systematic analysis of progress towards Millennium Development Goal 5. Lancet, 2010; 375(9726):1609–1623.

17.

Okigbo

, Speizer

, Corroon

, et al. Exposure to family planning messages and modern contraceptive use among men in urban Kenya, Nigeria, and Senegal: A cross-sectional study. Reprod Health, 2015; 12(1):63.

18.

Darroch

, Singh

. Trends in contraceptive need and use in developing countries in 2003, 2008, and 2012: An analysis of national surveys. Lancet, 2013; 381(9879):1756–1762.

19.

Dwomoh

, Amuasi

, Amoah

, et al. Exposure to family planning messages and contraceptive use among women of reproductive age in sub-Saharan Africa: A cross-sectional program impact evaluation study. Sci Rep, 2022; 12(1):18941.

20.

Tsui

, Brown

, Li

. Contraceptive practice in sub-Saharan Africa. Popul Dev Rev, 2017; 43(Suppl (Suppl 1)):166–191.

21.

Admasu

, Tamir

, Alemu

, et al. Prevalence and determinants of exposure to family planning messages among reproductive age women in five Sub-Saharan African countries: A multilevel analysis of 2021–2023 demographic and health surveys. BMC Womens Health, 2025; 25(1):365.

22.

Ajaero

, Odimegwu

, Ajaero

, et al. Access to mass media messages, and use of family planning in Nigeria: A spatio-demographic analysis from the 2013 DHS. BMC Public Health, 2016; 16(1):427.

23.

Sedgh

, Ashford

, Hussain

. Unmet need for contraception in developing countries: Examining women’s reasons for not using a method. 2016.

24.

Gupta

, Katende

, Bessinger

. Associations of mass media exposure with family planning attitudes and practices in Uganda. Stud Fam Plann, 2003; 34(1):19–31.

25.

Westoff

, Bankole

. Mass Media and Reproductive Behavior in Africa. Macro International Calverton, MD; 1997.

26.

Ezeh

, Kodzi

, Emina

. Reaching the urban poor with family planning services. Stud Fam Plann, 2010; 41(2):109–116.

27.

Bongaarts

. The impact of family planning programs on unmet need and demand for contraception. Stud Fam Plann, 2014; 45(2):247–262.

28.

Mutumba

. Mass media influences on family planning knowledge, attitudes and method choice among sexually active men in sub-Saharan Africa. PLoS One, 2022; 17(1):e0261068.

29.

Singh

, Singh

, Verma

. Knowledge, attitude and practice GAP in family planning usage: An analysis of selected cities of Uttar Pradesh. Contracept Reprod Med, 2016; 1(1):20.

30.

Speizer

, Corroon

, Calhoun

, et al. Demand generation activities and modern contraceptive use in urban areas of four countries: A longitudinal evaluation. Glob Health Sci Pract, 2014; 2(4):410–426.

31.

Alkema

, Kantorova

, Menozzi

, et al. National, regional, and global rates and trends in contraceptive prevalence and unmet need for family planning between 1990 and 2015: A systematic and comprehensive analysis. Lancet, 2013; 381(9878):1642–1652.

32.

McNab

, Freedman

. Maternal newborn health and the urban poor: A global scoping. New York Columbia Univ Mail Sch Public Heal. Published online 2016.

33.

SLDHS. The Somaliland Health and Demographic Survey 2020. Available from: https://www.somalilandcsd.org/somaliland-health-and-demography-survey-slhd2020/

34.

Hussein

, Abdi

, Yousuf

, et al. Spatial distribution and determinants of unimproved sources of drinking water among Somali households: Spatial and multilevel analysis using nationwide survey data. J Water Health, 2025:jwh2025140.

35.

Goldstein

. Multilevel Statistical Models. John Wiley & Sons; 2011.

36.

Moraga

. Spatial Statistics for Data Science: Theory and Practice with R. CRC Press; 2023.

37.