Factors associated with unmet need for family planning in sub-Saharan Africa: A multilevel multinomial logistic regression analysis.

BACKGROUND: More than one out of every ten married women in the world, and one out of every five women in Africa, have unmet family planning needs. Despite this, studies concerning sub-Saharan Africa as well as the community-level factors that may influence the unmet need for family planning are scarce.
OBJECTIVE: To assess factors associated with unmet need for family planning in sub-Saharan Africa.
METHODS: This study used the nineteen demographic and health surveys (DHS) conducted between 2015 and 2020 in sub-Saharan Africa. A total weighted sample of 175, 820 women of reproductive age who were married during the survey was used for this study. To assess the factors associated with unmet need for family planning, I have employed a multilevel multinomial logistic regression model. After selecting variables using the bivariable analysis, a multivariable model was fitted. Finally, an adjusted relative risk ratio with its 95% Confidence Interval was reported and variables with a p-value less than 0.05 were declared to be significant predictors of unmet need for family planning. RESULT: The overall prevalence of unmet need for family planning in sub-Saharan Africa was 23.70%, of which unmet need for spacing and limiting was 15.81% and 7.90% respectively. In the multivariable multilevel multinomial model, women's age, women's education, age at cohabitation, heard about family planning through media, parity, number of under-five children, and knowledge about modern contraceptive methods were among the individual-level factors that were associated with both the unmet need for spacing and limiting. Place of residence, community level of women illiteracy, and region were among the community-level factors that were associated with both unmet needs for spacing and limiting. Household size and visiting the health facility in the last 12 months were associated with unmet need for spacing only and husband education was associated with unmet need for limiting only.
CONCLUSION: Unmet need for family planning in sub-Saharan Africa was high. Both individual and community level factors were associated with both unmet need for spacing and limiting. Therefore, it is better to consider interventions at both individual and community levels.

Background

Even though reproductive and sexual health care services are very crucial, their availability to women of reproductive age is quite limited [1]. One of these services is family planning and a lack of family planning or not using it regularly or appropriately leads to unwanted pregnancy, which is one of the leading causes of maternal mortality [2]. Unintended pregnancy affects almost every one of life and is a huge and global public health issue. It is linked to parental stress, induced abortion, infertility, negative physical, mental, social, and economic consequences, as well as maternal and child mortality [1,3,4].

The most common cause of unwanted pregnancies among married women is an unmet demand for family planning [5,6]. It is the percentage of women who are currently married or in a partnership who decide to stop or delay pregnancy/childbearing but without using contraception [7-9].

In 2017, the majority (63%) of reproductive-age women who were married or in a partnership used contraception globally; however, there was significant variation between regions. More than one out of every ten married women in the world, and one out of every five women in Africa, have unmet family planning needs [10].

According to different pieces of literature, individual-level factors such as women age, women education, husband education, wealth index, media exposure, age at first marriage, number of living children, parity, household size, health care decision making, knowledge about family planning method and different community-level factors are related with such higher occurrence of unmet need for family planning [11-24].

Unmet family planning needs are an important topic to consider when lobbying for reproductive health, developing family planning policy, and monitoring and assessing family planning programs. Despite the fact, prior studies have addressed the unmet need for family planning, most of them focus on a single country, and many of them ignore community-level factors that may influence the unmet need for family planning. Therefore, this study aimed to assess factors associated with unmet need for family planning among married women of reproductive age in sub-Saharan Africa using a multilevel multinomial logistic regression analysis.

Methods

Data source and study population

This study used the nineteen demographic and health surveys (DHS) conducted between 2015 and 2020 in sub-Saharan Africa. The DHS is a survey conducted every five years. It is based on a two-stage sampling technique. In the first stage, clusters/enumeration areas were selected using each country’s most recent population and housing census as a sampling frame, and in the second stage; households were sampled from the newly created household listing. The survey target groups were women and men of reproductive age in randomly selected households of each country. The detailed information was collected on background characteristics, maternal and child health, HIV/AIDS, domestic violence, family planning services, and other important public health problems. For this study, we have appended women’s data of each country. Only women of reproductive age who were married were included since they are exposed to regular sexual intercourse. Nineteen countries’ DHS data with a total weighted sample of 175, 820 women of reproductive age who were married during the survey was used for this study (Table 1). Further information regarding the DHS methodology can be found elsewhere [25].

Table 1

Sample size in each country, study year, and total sample size.

Country	Year	Total population (Weighted)	Frequency
Angola	2015	7,957	4.53
Benin	2017/18	11,168	6.35
Burundi	2016/17	9,782	5.56
Cameroon	2018/19	7,748	4.41
Ethiopia	2016	10,223	5.81
Gambia	2019/20	7,526	4.28
Guinea	2018	7,727	4.39
Liberia	2019/20	4,216	2.40
Mali	2018	8,567	4.87
Nigeria	2018	29,090	16.55
Rwanda	2015	6,978	3.97
Sierra Leone	2019	9,715	5.53
Senegal	2019	5,657	3.22
Chad	2015	13,184	7.50
Tanzania	2015/16	8,210	4.67
Uganda	2016	11,223	6.38
South Africa	2016	3,050	1.73
Zambia	2018/19	7,648	4.35
Zimbabwe	2015	6,151	3.50
Total	2015–2020	175,820	100.00

Variables of the study

Outcome variable

Unmet need for family planning was the dependent variable. It had three categories; the unmet need for spacing (women who wanted to wait/delay in having another child, but not using any form of contraception), unmet need for limiting (women who did not want any more children, but did not use any contraceptive method), and no unmet need. This classification was made using the DHS’s recently revised definition of unmet need for family planning [26].

Explanatory variables

Both individual and community level variables were incorporated as explanatory variables after reviewing the DHS data.

Individual-level variables. Women age, women and husbands education, wealth index, age at cohabitation, health care decision making, visited a health facility in the last 12 months, hearing about family planning methods by television, radio, or/and newspaper/magazines, parity, number of under-five children, household size, and having knowledge on family planning methods were the individual-level variables.

Community-level variables. The community-level variables incorporated for this study were; residence, African region, community level of women illiteracy, community level of media exposure, community level of child care burden, and community poverty level. Except for residence and African region, all other community-level variables were created by aggregating individual-level variables into cluster/community level variables as follows:

Community-level of women illiteracy; the proportion of women with no formal education derived from data on women’s level of education and categorized as low and high community level of women education (using national median value).

Community-level childcare burden; the proportion of women who had five or more under-five children derived from data on the total number of under-five children and categorized as low and high community level of childcare burden based on a national median value.

Community poverty level; the proportion of women in the poorest and poorer quintiles derived from data on wealth index and categorized as low and high poverty communities based on a national median value.

Community-level media exposure; was measured by the proportion of women who had been exposed to family planning methods through at least one media (either television, radio, and/or newspaper/magazine) and categorized as low and high community level of media exposure like the above community-level variables.

Data management and statistical analysis

The data was appended, recoded, and analyzed using Stata version 16 software. To restore representativeness and to get an appropriate statistical estimate, every analysis was based on weighting. Both descriptive (frequencies and percentages) and analytical analysis (multilevel multinomial logistic regression) were done and reported.

Model building

I have employed a multilevel multinomial logistic regression analysis, which includes both random effect and fixed-effect models. This statistical approach was done in Stata version 16 software via a Generalized Structural Equation Modelling (with the logit link function) using the gsem Stata command. Four different nested models were fitted. The models were null model (containing only the outcome variable), model 1 (a model fitted using individual-level variables only), model 2 (a model fitted using community-level variables), and model 3 (fitted using both individual and community level variables). The final interpretation of results were based on the best model among them, which is selected through the use of the log-likelihood and Akaike’s information criteria. Both bivariable (to select eligible variables for the multivariable analysis) and multivariable analysis was employed. P-value less than 0.20 was used to retain and include variables in the multivariable analysis. Finally, the adjusted Relative Risk Ratio (RRR) with 95% Confidence Interval (CI) was reported and variables with p value<0.05, in the multivariable analysis, were declared to be significant predictors of unmet need for family planning (unmet need for spacing and limiting). Furthermore, I have checked Multicollinearity between explanatory variables using Variance Inflation Factor (VIF) and found that there was no Multicollinearity (the VIF ranges from 1.03 to 2.09 with the mean VIF of 1.45)

In the random-effects analysis, to assess the variability of unmet need for family planning between clusters/communities, both Intraclass Correlation Coefficient (ICC) and Proportional change in Variance (PCV) were calculated.

Ethical consideration

Since I have used the publicly available data, ethical approval was not required. However, I have accessed the data set from the DHS website (https://dhsprogram.com) after telling the purpose of the study.

Results

Sociodemographic characteristics of respondents

A total weighted sample of 175,820 women of reproductive age who were married was used for the final analysis. Most (16.55%) of the study participants were from Nigeria (Table 1). The median age of the study participants was 30 (IQR = 25–38) years. The majority (43.48%) had no formal education and only 4.64% had higher education. More than half (52.67%) of the respondents were married at the age of 18 and older. Around 58.94% of the respondents visit the health facility during the last 12 months. The majority of the respondents did not heard about family planning methods through radio (66.40%), television (83.54%), and newspaper/magazine (93.69%). More than half (54.57%) of the respondents were from a household size of five and above and 93.45% of the respondents know about modern family planning methods. Regarding the place of residence and region, 64.91% and 47.59% of the respondents were from rural residence and western African region respectively (Table 2).

Table 2

Sociodemographic characteristics of respondents and their partner.

Characteristics	Frequency (N = 175,820)	Percentage (%)
Maternal age    15–19    20–24    25–29    30–34    35–39    40–44    45–49	11,80729,07337,94532,97328,37319,98915,660	6.2716.5421.5818.7516.1411.378.91
Wealth index    Poorest    Poorer    Middle    Rich    Richer	35,08536,15334,90035,05934,623	19.9520.5619.8519.9419.69
Women education    No formal education    Primary    Secondary    Higher	76,88151,24839,5368,155	43.7329.1522.494.64
Husband education    No formal education    Primary    Secondary    Higher	69,15545,95145,89414,820	39.3326.1426.108.43
Age at cohabitation (years)    <18    ≥18	83,22292,598	47.3352.67
Health care decision    Women only    Both    Husband only    Others*	28,69269,49376,4541,181	16.3239.5343.480.67
Visited Health facility during the last 12 months    No    Yes	72,184103,636	41.0658.94
Heard about family planning through the radio last few months    No    Yes	116,74359,077	66.4033.60
Heard family planning in newspaper magazine last few months    No    Yes	164,73011,090	93.696.31
Heard family planning on television last few months    No    Yes	146,87428,946	83.5416.46
Parity    0    1    2–4    5 and above	11,61824,78577,39162,026	6.6114.1044.0235.28
Number of under 5 children    None    One    Two    Three and above	36,80558,54649,10031,369	20.9333.3027.9317.84
Household size    1–2    3–4    5 and above	9,66270,20695,952	5.5039.9354.57
Having knowledge about modern family planning methods    No    Yes	11,509164,311	6.5593.45
Residence    Urban    Rural	61,692114,128	35.0964.91
Community-level of women illiteracy    Low    High	88,95586,865	50.5949.41
Community-level media exposure    Low    high	87,38988,431	49.7050.30
Community-level child care burden    Low    high	87,78188,039	49.9350.07
Community poverty level    Low    High	91,43784,383	52.0147.99
African region    Eastern    Western    Central    Southern	60,21583,66628,8893,050	34.2547.5916.431.73

Note

* = someone else and other individuals.

Unmet need for family planning in sub-Saharan Africa

The overall prevalence of unmet need for family planning was 23.70% (95%CI; 23.50, 23.90). The prevalence of unmet need for spacing and limiting was 15.81% (95%CI; 15.64, 15.98) and 7.90% (95%CI; 7.77, 8.02) respectively.

Factors associated with unmet need for family planning in sub-Saharan Africa

A multilevel multinomial logistic regression analysis was fitted to assess the factors associated with unmet need for family planning. In this study, in the multivariable multilevel multinomial regression analysis, both individual level and community level variables were associated with both unmet needs for spacing and limiting.

Factors associated with unmet need for spacing

Women age, women education, age at cohabitation, health care decision making, visiting health facility in the last 12 months, heard about family planning through reading newspaper/magazine and watching television, parity, number of under-five children, household size, knowledge about modern contraceptive methods were among the individual-level factors that were associated with unmet need for spacing. Place of residence, community level of women illiteracy, and region were among the community-level factors that were associated with unmet need for spacing. Being in an older age group was associated with a lower risk of having an unmet need for spacing compared with younger-aged women. Women with primary education and higher education had 10% (RRR = 1.10, 95%CI; 1.05, 1.15) higher risk and 12% (RRR = 0.88, 95%CI; 0.78, 0.99) lower risks of having an unmet need for spacing, respectively. Women who have got married at or above the age of 18 had 32% (RRR = 1.32, 95%CI; 1.27, 1.37) higher risk of having an unmet need for spacing as compared to their counterparts. Regarding health care decision, a decision made by both women and her husband, husband only, and others had 6% (RRR = 0.96, 95%CI; 0.90, 0.98) lower risk, 5% (RRR = 0.95, 95%CI; 0.90, 0.99) lower risk, and 24% (RRR = 1.24, 95%CI; 1.04, 1.49) higher risk of having an unmet need for spacing as compared to decisions made by women only respectively. Being visiting the health facility in the last 12 months was associated with 7% (RRR = 1.07, 95%CI; 1.04, 1.11) higher risks of unmet need for spacing as compared to their counterparts. Being heard about family planning through television and newspaper/magazine were associated with 11% (both have similar figure) lower risks of having an unmet need for spacing respectively. Higher parity, a higher number of under-five children, being from the nuclear family, and knowing modern contraceptive methods were also associated with higher risks of having an unmet need for spacing. Being from rural residence (RRR = 0.86, 95%CI; 0.82, 0.91) and communities with higher illiteracy level (RRR = 0.94, 95%CI; 0.90, 0.99) were associated with lower risks of having an unmet need for spacing as compared to their counterparts respectively. Moreover, being from the western African region (RRR = 1.25, 95%CI; 1.19, 1.33), central African region (RRR = 1.48, 95%CI; 1.39, 1.57), and southern African region (RRR = 0.71, 95%CI; 0.57, 0.89) were associated with having an unmet need for spacing (Table 3).

Table 3

Multilevel multinomial logistic regression analysis in assessing the factors associated with unmet need for family planning for (both spacing and limiting).

Characteristics	Null model	Model 1		Model 2		Model 3
		Individual-level characteristics		Community-level characteristics		Both individual and community-level characteristics
		Unmet need for spacingRRR (95%CI)	Unmet need for limitingRRR (95%CI)	Unmet need for spacingRRR (95%CI)	Unmet need for limitingRRR (95%CI)	Unmet need for spacingRRR (95%CI)	Unmet need for limitingRRR (95%CI)
Maternal age    15–19    20–24    25–29    30–34    35–39    40–44    45–49		1.000.65 (0.61, 0.70)0.50 (0.46, 0.54)0.41 (0.38, 0.45)0.31 (0.28, 0.34)0.18 (0.16, 0.20)0.07 (0.06, 0.08)	1.000.69 (0.53, 0.91)0.87 (0.67, 1.13)1.57 (1.2, 2.04)2.55 (1.96, 3.31)3.43 (2.6, 4.49)2.55 (1.95, 3.34)			1.000.66 (0.62, 0.71)***0.51 (0.47, 0.55)***0.42 (0.39, 0.46)***0.32 (0.29, 0.35)***0.18 (0.16, 0.20)***0.07 (0.06, 0.08)***	1.000.68 (0.52, 0.90)**0.85 (0.65, 1.10)1.51 (1.16, 1.96)**2.45 (1.87, 3.16)***3.27 (2.49, 4.28)***2.43 (1.86, 3.18)***
Wealth index    Poorest    Poorer    Middle    Rich    Richer		1.001.03 (0.98, 1.08)1.06 (1.01, 1.12)1.06 (1.00, 1.13)1.08 (1.01, 1.16)	1.001.05 (0.98, 1.13)1.02 (0.95, 1.09)1.07 (0.99, 1.15)1.14 (1.04, 1.24)			1.001.01 (0.96, 1.06)1.01 (0.96, 1.06)0.98 (0.92, 1.04)0.95 (0.88, 1.03)	1.001.05 (0.98, 1.12)0.99 (0.93, 1.07)1.01 (0.94, 1.09)1.03 (0.94, 1.13)
Women education    No formal education    Primary    Secondary    Higher		1.001.06 (1.02, 1.11)0.98 (0.93, 1.04)0.89 (0.79, 1.01)	1.001.10 (1.04, 1.17)1.05 (0.96, 1.15)0.89 (0.75, 1.05)			1.001.10 (1.05, 1.15)***0.97 (0.92, 1.02)0.88 (0.78, 0.99)*	1.001.07 (1.00, 1.13)*0.98 (0.90, 1.08)0.84 (0.71, 0.99)*
Husband education    No formal education    Primary    Secondary    Higher		1.000.93 (0.89, 0.98)0.97 (0.92, 1.02)0.96 (0.88, 1.04)	1.001.11 (1.05, 1.18)1.12 (1.04, 1.21)1.07 (0.95, 1.20)			1.000.99 (0.94, 1.0400.96 (0.91, 1.01)0.95 (0.87, 1.03)	1.001.06 (1.02, 1.16)*1.10 (1.01, 1.18)*1.06 (0.94, 1.20)
Age at cohabitation (years)    <18    ≥18		1.001.30 (1.25, 1.35)	1.001.07 (1.01, 1.12)			1.001.32 (1.27, 1.37)***	1.001.06 (1.01, 1.11)*
Health care decision    Women only    Both    Husband only    Others		1.000.95 (0.90, 0.98)1.01 (0.96, 1.06)1.31 (1.09, 1.56)	1.000.81 (0.76, 0.86)0.68 (0.64, 0.73)0.80 (0.58, 1.10)			1.000.94 (0.90, 0.98)**0.95 (0.90, 0.99)*1.24 (1.04, 1.49)*	1.000.82 (0.77, 0.87)***0.71 (0.67, 0.76)***0.83 (0.60, 1.14)
Visited Health facility during the last 12 months    No    Yes		1.001.04 (1.01, 1.08)	1.001.01 (0.96, 1.06)			1.001.07 (1.04, 1.11)***	1.000.99 (0.94, 1.04)
Heard about FP through the radio last few months    No    Yes		1.001.00 (0.96, 1.04)	1.000.95 (0.89, 1.01)			1.001.04 (0.99, 1.08)	1.000.95 (0.90, 1.01)
Heard family planning in newspaper magazine last few months    No    Yes		1.000.84 (0.77, 0.93)	1.000.86 (0.76, 0.97)			1.000.89 (0.80, 0.96)***	1.000.83 (0.73, 0.93)**
Heard FP on television last few months    No    Yes		1.000.95 (0.89, 1.01)	1.000.02 (0.94, 1.11)			1.000.89 (0.84, 0.95) **	1.001.01 (0.94, 1.10)
Parity    0    1    2–4    5 and above		1.001.86 (1.70, 2.05)2.21 (2.01, 2.44)2.97 (2.67, 3.31)	1.001.71 (1.11, 2.64)6.42 (4.41, 9.34)15.61 (10.69, 22.81)			1.001.85 (1.68, 2.04)***2.18 (1.98, 2.40)***3.90 (2.60, 3.23)***	1.001.69 (1.10, 2.61)*6.37 (4.38, 9.27)***15.88 (10.87, 23.19)***
Number of under 5 children    None    One    Two    Three and above		1.002.33 (2.16, 2.52)3.13 (2.89, 3.38)3.62 (3.32, 3.95)	1.002.16 (1.99, 2.34)2.34 (2.16, 2.53)2.27 (2.08, 2.48)			1.002.34 (2.16, 2.52)***3.10 (2.86, 3.36)***3.50 (2.22, 3.83)***	1.002.17 (2.00, 2.35)***2.36 (2.18, 2.56)***2.33 (2.14, 2.55)***
Household size    1–2    3–4    5 and above		1.000.80 (0.72, 0.89)0.78 (0.70, 0.88)	1.000.83 (0.69, 1.00)0.88 (0.74, 1.06)			1.000.79 (0.71, 0.88)***0.77 (0.68, 0.86)***	1.000.85 (0.71, 1.02)0.91 (0.76, 1.09)
Having knowledge about modern family planning methods    No    Yes		1.001.00 (0.93, 1.07)	1.001.33 (1.19, 1.48)			1.001.10 (1.02, 1.89)*	1.001.26 (1.12, 1.41)***
Community-level characteristics of respondents
Residence    Urban    Rural				1.001.04 (0.99, 1.08)	1.001.02 (0.97, 1.08)	1.000.86 (0.82, 0.91)***	1.000.83 (0.77, 0.89)***
Community-level of women illiteracy    Low    high				1.000.97 (0.93, 1.02)	1.000.88 (0.83, 0.93)	1.000.94 (0.90, 0.99)*	1.000.87 (0.82, 0 .92)***
Community-level media exposure    Low    high				1.000.97 (0.93, 1.01)	1.000.95 (0.89, 1.01)	1.000.98 (0.94, 1.03)	1.000.95 (0.90, 1.01)
Community-level child care burden    Low    high				1.001.05 (1.01, 1.10)	1.001.00 (0.94, 1.05)	1.000.98 (0.94, 1.03)	1.000.97 (0.92, 1.03)
Community poverty level    Low    High				1.001.01 (0.96, 1.06)	1.000.99 (0.93, 1.05)	1.000.97 (0.93, 1.02)	1.000.97 (0.91, 1.03)
African region    Eastern    Western    Central    Southern				1.001.23 (1.18, 1.29)1.56 (1.48, 1.65)0.42 (0.34, 0.53)	1.000.85 (0.81, 0.90)0.86 (0.80, 0.94)0.93 (0.78, 1.11)	1.001.25 (1.19, 1.33)***1.48 (1.39, 1.57)***0.71 (0.57, 0.89)**	1.000.87 (0.82, 0 .94)**0.85 (0.77, 0.94)**1.47 (1.21, 1.80)**

Note

* = p<0.05

** = p<0.01

*** = p<0.001.

Factors associated with unmet need for limiting

Women age, women education, husband education, age at cohabitation, health care decision making, heard about family planning methods through magazine or newspaper, parity, number of under-five children, and knowledge about modern contraceptive methods were the individual-level factors that were associated with unmet need for limiting. While place of residence, community level of women illiteracy, and region were among the community-level factors that were associated with unmet need for limiting. Women in the age group 20–24 had 32% (RRR = 0.68, 95%CI; 0.52, 0.90) lower risks for having an unmet need for spacing. While women in the age group 30–34, 35–39, 40–44, and 45–49 had 1.51 (RRR = 1.51, 95%CI; 1.16, 1.96), 2.45 (RRR = 2.45, 95%CI; 1.87, 3.16), 3.27 (RRR = 3.27, 95%CI; 2.49, 4.28), and 2.43 (RRR = 2.43, 95%CI; 1.86, 3.18) times higher risk of having unmet need for limiting respectively. Having primary education is associated with a higher risk of having an unmet need for limiting and being having a higher education is associated with a lower risk of having an unmet need for limiting. Women with their husbands having primary education and secondary education had a higher risk of having an unmet need for limiting as compared to those women with their husbands not having a formal education. Women who have married at the age of 18 and above had a 1.06 (RRR = 1.06, 95%CI; 1.01, 1.11) times higher risk of having an unmet need for limiting. Regarding health care decision making, making a decision with their husband and by their husband only is associated with lower risks of having an unmet need for limiting. Being hearing about family planning through newspapers or magazines was associated with 17% (RRR = 0.83, 95%CI; 0.73, 0.93) lower risks of having an unmet need for limiting. Women with parity of one, 2–4, and 5 and above had 1.69 (RRR = 1.69, 95%CI; 1.10, 2.61), 6.37 (RRR = 1.37, 95%CI; 4.38, 9.27), and 15.88 (RRR = 15.88, 95%CI; 10.87, 23.19) times higher risks of having an unmet need for limiting as compared to nulliparous women. Having one, two, and three and above under-five children is associated with 2.17 (RRR = 2.17, 95%CI; 2.00, 2.35), 2.36 (RRR = 2.36, 95%CI; 2.18, 2.56), and 2.33 (RRR = 2.33, 95%CI; 2.14, 2.55) times higher risks of having an unmet need for limiting respectively as compared to those with no under-five children. Women who knew about family planning have 1.26 (RRR = 1.26, 95%CI; 2.14, 2.55) times higher risks of having an unmet need for limiting as compared to their counterparts. Moreover, women from rural areas and communities with higher women illiteracy level had 17% (RRR = 0.83, 95%CI; 0.77, 0.89) and 13% (RRR = 0.87, 95%CI; 0.82, 0.92) lower risks of having an unmet need for limiting. There was also regional variation in the unmet need for limiting (Table 3).

Random effect analysis

As shown in Table 4, the ICC in the null model revealed that about 14% of the variability in unmet need for family planning was attributed due to differences between clusters/communities. While the highest PCV in the final model revealed that about 22% of the variability of unmet need for family planning was explained by both individual and community-level factors. Regarding model fitness, the model with the highest log-likelihood and the lowest Akaike’s information criteria, model 3, was the best-fitted model (Table 4).

Table 4

The random effect model in assessing the factors associated with unmet need for family planning in sub-Saharan Africa.

Parameter	Null model	Model 1	Model 2	Model 3
Community level variance	0.520	0.500	0.494	0.406
Intraclass Correlation Coefficient	0.140	0.132	0.130	0.110
Proportional Change in Variance	Reference	0.038	0.050	0.220
Model fitness
Log-likelihood	-122553	-110832.3	-122070.7	-110512
Akaike’s information criteria	245111.9	221802.5	244179.4	221194.1
Bayesian information criteria	245142.1	222497.9	244370.8	222050.7

Discussion

This study aimed to assess factors associated with unmet need for family planning in sub-Saharan Africa. The study at hand identified both individual and community levels factors that are associated with the unmet need for limiting and spacing.

This study revealed that the unmet need for spacing is decreased as a woman goes from the age group 15–19 to the next consecutive age groups (older age groups). Regarding the unmet need for limiting, it is lower as women go from 15 to 19 to 20–24. However, the unmet need for limiting is higher as the age group gets old than 24. This is in line with studies conducted in Ghana, Ethiopia, and Kenya [27-30]. This is since younger-aged women would prefer short-term family planning methods to space their next births and are more likely to have an unmet need for spacing. Besides, this may be because younger women are less likely to have achieved their fertility goals and therefore, want to space than limit the number of births. However, older aged women may have an ideal family size (may have achieved their fertility goals) and tend to prefer long-term contraceptive methods and in some instances permanent methods of contraception and, therefore, they are more likely to have an unmet need for limiting.

This study revealed a higher unmet need for both limiting and spacing among respondents who had completed primary education and among women from communities with higher women illiteracy level. This is in line with study findings from Ghana and Ethiopia [27,31,32]. This could be because women with only a primary education and those from low-literacy communities may unaware of where contraceptive methods can be found and when to use them. However, in this study, women with higher educational level had a lower risk for both unmet needs for limiting and spacing as compared to those without formal education. This is in line with studies conducted elsewhere [29,32,33]. The likely explanation is that, unlike women with no formal education, educated women do not have deep rooted misconceptions about the benefits of contraception and tend to have strong positive attitudes regarding family planning methods [28]. Meanwhile, women with higher partner’s education (primary and secondary education as compared with not having formal education) had a higher tendency to experience the unmet need for limiting but not associated with unmet need for spacing.

The study at hand also revealed that having knowledge about modern family planning methods is associated with both unmet needs for spacing and limiting. That means women who had knowledge about family planning have a higher unmet need for spacing and limiting. This is inconsistent with study findings from Ethiopia [34] and Nigeria [35]. The likely reason is that women who are more knowledgeable about family planning methods are more aware of the potential risks, as a result, they may have an increased unmet need due to fear of side effects. However, this is an unusual finding and it needs further investigation.

Those who heard about family planning through media (television and newspaper/magazine) were less likely to experience an unmet need for family planning. This is in agreement with study findings from Ethiopia [30,32] and Pakistan [33]. The explanation for this could be that obtaining information about family planning methods through various media can assist women in learning about and understanding family planning options that are accessible and available.

In this study, age at marriage had associated with the unmet need to limit and space births. That is women who got married at the age of 18 or older were more likely to have a higher level of unmet need for spacing and limiting. This finding is consistent with other studies conducted elsewhere [29,30,36]. This is because, compared to women who were married before the age of 18, those who married after the age of 18 are more prone to accept attitudes about birth restrictions and, therefore, more likely to have higher demand and unmet need for family planning [15].

The study at hand revealed that the involvement of the husband in women’s health care decisions decreases the unmet need for limiting and spacing. This finding is consistent with a study conducted in Burkina Faso [37], which revealed that husbands’ approval of health care such as family planning is a protective factor of unmet needs. This may be because the involvement of a husband in health care decision-making, especially decisions about family planning, helps the woman to have an open discussion with her partner and make an appropriate decisions for their health and this may result in a lower unmet need for family planning.

Women with higher numbers of under-five children are more likely to have unmet needs for both limiting and spacing of births than women with no under-five children. This finding is in line with study results from Kenya [28] and Ethiopia [15,30]. In addition, in this study, women with higher parity are more likely to have an unmet need for both spacing and limiting childbirth. This is because these group of women may attain their ideal family sizes (required numbers of children) and hence are more likely to have high unmet need to limit the birth of additional children and/or to delay the next childbirth.

Moreover, in this study, women from higher numbers of households compared to those with one to two household members had higher risks for unmet need for spacing but it is not associated with unmet need for limiting. This finding is agreed with a study done in Afghanistan [19]. This could be because mothers of big families are preoccupied with caring for their family and are unable to seek health care, including family planning [38,39].

In this study, place of residence had significantly associated with both unmet needs for spacing and limiting. Rural dwellers were less likely to report unmet needs for limiting and spacing. This is consistent with a study finding from Ghana [40] and inconsistent with prior studies conducted elsewhere [31,33]. The possible explanation is that rural women have relatively lower decision-making power about family planning use and could not report contraceptive use levels and have lower reports in unmet needs. However, the author recommends further studies in this regard.

The study has both strengths and limitations. It is useful for policymakers and program planners to create intervention strategies because it is a multi-country analysis based on nationally representative data. In addition, for better parameter estimation, we used a multilevel multinomial logistic regression approach. However, investigating causality between dependent and independent factors is difficult due to the cross-sectional nature of the data. Furthermore, because it is reliant on available factors in the data set, the most relevant explanatory variables may be overlooked.

Conclusion

In this study, the overall magnitude of unmet need for family planning was high. Both individual and community-level factors were associated with both unmet needs for family planning (unmet need for spacing and limiting). Therefore, it is better to give special attention to those women who are at risk of having a higher unmet need for family planning. Moreover, media campaigns related to family planning will have a great impact to reduce the unmet need for family planning.

12 Jan 2022

Submitted filename: unmet need for FP comments.docx

Click here for additional data file.

20 Jan 2022

Date: January 20, 2022

Point by point response

Title: Factors associated with unmet need for family planning in sub-Saharan Africa: A multinomial multilevel logistic regression analysis

Manuscript number: PONE-D-21-28601

Dear editor and reviewers, thank you for the comments and suggestions you gave for this manuscript. I found all the comments and suggestions very constructive and important for the improvement the manuscript. Here, below, is the point by point response for your comments and suggestions, besides, I have incorporated all these in the revised manuscript.

Reviewer #1

The author is expected to state whether or not there is used/followed guideline such as strengthening the Reporting of Observational Studies in Epidemiology (STROBE) or others to conduct this study and write the manuscript. Using such guidelines will enhance the validity of the report.

Author’s response: Thank you very much. I have followed the STROBE guideline/checklist while writing this manuscript, see the revised manuscript.

Abstract (background)- the author wrote the facts (the figure) regarding the unmet need for FP. However, the author did not state the research gap. Therefore, reconsider it and add a sentence that indicates the research gap.

Author’s response: Thank you. The research gap is incorporated in the revised manuscript, see the abstract (background).

Abstract (conclusion)- the conclusion should be in line with the objective of the study. In this case, the author has two major objectives as a research question (i.e., magnitude and factors associated with unmet need). Therefore, the author has to conclude to those major research questions.

Author’s response: Thank you. The conclusion is amended, it is based on the objectives, in the revised manuscript based on the comment you gave me.

Explanatory variables-the author has indicated that both individual and community-level variables were identified after searching different kinds of literature. However, since the source of data is DHS, incorporating explanatory variables from other sources is impossible as we have to only rely on the variables that can be obtained from this secondary data. Therefore, re-write the sentence as: “Both individual and community level variables were incorporated as explanatory variables after reviewing the DHS data”.

Author’s response: Thank you for your comment and recommendation. I have modified the sentence to read “Both individual and community level variables were incorporated as explanatory variables after reviewing the DHS data”.

Data management and statistical analysis- the author used so many variables for the multivariable analysis. As a result, the chance of having multicollinear variables will be increase. Therefore, the author has to report information regarding multicollinearity and the used technique (s) to diagnose it.

Author’s response: Thank you very much for raising such very important issue. I have checked Multicollinearity using VIF test and there was no Multicollinearity. That is the VIF ranges from 1.03 to 2.09 with the mean VIF of 1.45, see the data management and statistical analysis section in the revised manuscript.

Ethical consideration- specify the date when the data were requested and accessed from the DHS program.

Author’s response: Thank you. I have accessed the data set on July 2021 within two days of asking the Measure DHS program at https://dhsprogram.com.

Discussion- reduce the first paragraph only to the objective of the study. Other points written here seem results. Hence, move and reshuffle it under the result section.

Author’s response: Thank you for the comment. I have reduced the first paragraph the discussion section accordingly.

Discussion- the author has put justification for every comparable or contrasting result. This is very appreciable. However, the justification sentences have to be short and precise. Therefore, chop those sentences into short form. Also, put references for some of your justifications.

Author’s response: Thank you. I have considered the issue in the revised manuscript.

Conclusion- the author put the findings (i.e., the prevalence of unmet need for FP and its correlates) of the study as a conclusion. However, the conclusion needs to have a public health implication which is not stated in this manuscript. Public health implication means what does the prevalence indicate according to the current standards of service (national/global target, plan, etc.)? Again, does your finding is low or high according to the current standards of service? Therefore, the author has to forward a valid conclusion based on the actual findings.

Author’s response: Thank you. I have modified the conclusion accordingly in the revised manuscript

Reviewer #2:

Abstract

• Line 26, the author uses “we” as if they are more than one author on the paper. This should be looked at and corrected.

• Lines 41 and 42, author uses the words “community and regional” interchangeably, but these words can mean different descriptions. I suggest the author stick to the word “community”.

Author’s response: Thank you. I have considered your comment in the revised manuscript, see the revised abstract.

Background

• Line 55, the word “terminate” may not have been appropriately use here given the fact that terminate also denotes “abortion”. The author should kindly replace it with a more appropriate word.

Author’s response: I have modified the word to read “stop”

Methods

Explanatory variables

• The author talks about married or in a union in lines 83-85 but switch to age at cohabitation in lines 100-101. I suggest the author should kindly be consistent with the married or in a union throughout the manuscript.

Author’s response: Thank you. I have incorporated married women for the analysis and I rewrite the sentences accordingly in the revised manuscript.

Does “community level of women illiteracy” in line 106 mean the same as “Community-level of women education” in line 110? The author should kindly be consistent with one term throughout the manuscript.

Author’s response: Thank you. Community level of women illiteracy was consistently used in the revised manuscript.

• Lines 107-109 “Except for residence and African region, all other community-level variables were created by aggregating individual-level variables into cluster/community level variables as follows”.

Inferring from the above sentence, there is a high chance that some of the individual level variables will be correlated with the community level variables since the community level variables were derived from those individual level variables. For example, wealth index and community poverty level might be correlated, and the rest. Therefore, multicollinearity might have been an issue in the final model (model 3) where all these variables were modeled together. The author should check multicollinearity and report appropriately.

Author’s response: Thank you. I have checked Multicollinearity using VIF test and there was no Multicollinearity (the VIF ranges from 1.03 to 2.09 with the mean VIF of 1.45). I have incorporated the information showing this in the revised manuscript (see the data management and analysis section).

Line 128, “we” should kindly be replaced.

Author’s response: Corrected in the revised manuscript.

Results

• There are several grammar errors in the results section

Author’s response: I have checked the result section and grammatical errors are corrected

• Table 2, maternal age categorization is erroneous. The author should kindly correct it.

Author’s response: Amended in the revised manuscript.

• Table 2 and 3, the variable “health care decision” has a category referred to as “others”. This category was significantly associated with unmet need for spacing in the final model. Could the author explain what is meant by “others”? This may go to the Explanatory variable section under methods.

Author’s response: It was to mean a person other than the husband and themselves. In includes categories “someone else” and “other individuals” in the survey data and for this study, I have merged these in to “other”. I have clarified this in table 2, as a footnote.

• Line 153, add the “%” to the 43.48.

Author’s response: Corrected

• Lines 166-157, “The majority of the respondents had not heard about family planning methods through radio, television, and newspaper/magazine, respectively”. The author should kindly indicate the numbers in the above sentence.

Author’s response: The sentence is amended accordingly in the revised manuscript.

• Lines 158-159, does 93.45% constitute almost all?

Author’s response: Thank you. This was to indicate the figure/number is very high. But, in the revised manuscript, I have modified the sentence and the word “almost all” is deleted.

• Line 173-174, refer to my earlier in lines 83-85.

Author’s response: Corrected

• Lines 181-183 looks confusing. The author should kindly report RRR in the same direction if they are in the same sentence. This makes it clearer for readers. For example, primary education and higher education had 1.10 higher risk and 0.88 lower risk respectively or 10% higher risk and 12% lower risks. This may be applied appropriately throughout the results section.

Author’s response: Thank you. I have considered your suggestion/recommendation in the revised manuscript.

Discussion

• There are few grammar error here

Author’s response: Thank you. I have go through the overall manuscript and I have corrected grammatical errors.

• Lines 272-275, “This might be due to women with primary education did not make a wise decision regarding whether they will use contraceptives. Besides, women with a primary level of education may have a higher desire to postpone childbearing to fulfill their goals”.

• What does the author mean by “did not make a wise decision”? This might have been inappropriate to used.

• Besides, if a woman is 45 years old and her education level primary, how will this person postpone childbearing to fulfill her goals? This is not convincing enough, and I suggest the author think about other concrete explanation to this finding.

Author’s response: I have amended the sentence in the revised manuscript.

• Lines 280-282, the author should kindly reference any study that reports of health professionals having less unmet need for family planning compared to other professionals.

Author’s response: I have considered the comment in the revised manuscript, the justification is amended in the revised manuscript.

• Lines 302-304, the author should provide a reference for the explanation made.

Author’s response: Reference is added.

• Lines 312-318 sounds somehow contradictory. If women with higher number of under 5 years children and women with higher parity are more likely to have an unmet need for spacing and limiting, then how come they are also more likely to use contraceptives as stated in line 317-318?

Author’s response: Thank you. It was to mean more likely to have higher demand for family planning and I have modified the sentence in the revised manuscript.

• Lines 322-323, what about those who argue that mothers of big families rather have support systems such as in-laws and other siblings, and therefore might have more time to do other things including seeking for healthcare?

Author’s response: In Ethiopia, most of the mothers with big family are burdened by routine activities (child care, cooking, etc…). Most of them, carry the whole family. Therefore, they gave health care as secondary. And I have putted references for this justification in the revised manuscript.

Conclusion

• Radio, newspaper magazine, and television were the media outlets used by the author in this study. Interestingly, radio which is supposed to be widely accessible in countries involved in this study was not significant for both spacing and limiting unmet need on the final model. What then informed the author’s decision to suggest that media campaign would have a great impact to reduce family planning unmet needs?

Author’s response: Thank you. Even though I had expected Listening radio as associated factor for both unmet need for spacing and limiting, it was not the case in this study. However, the two media (radio and newspaper/magazine) were associated with unmet need for family planning and when I said media campaign, in the conclusion, it is to mean television and newspaper/magazine.

Submitted filename: Point by point response.docx

Click here for additional data file.

31 Jan 2022

Factors associated with unmet need for family planning in sub-Saharan Africa: A multilevel multinomial logistic regression analysis

PONE-D-21-28601R1

Dear Dr. Teshale,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Kannan Navaneetham, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

3 Feb 2022

PONE-D-21-28601R1

Factors associated with unmet need for family planning in sub-Saharan Africa: A multilevel multinomial logistic regression analysis

Dear Dr. Teshale:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Prof. Kannan Navaneetham

Academic Editor

PLOS ONE