Trends and determinants of contraceptive method choice

Studies show a gap in addressing the reproductive health Background: goals of younger women whose inconsistent use of contraception is high in spite of their great need for it. The women aged 15-24 present high potential for unintended pregnancy and increase the challenge for retention of users which is key in maintaining and pushing up the current gains in contraceptive prevalence rate (CPR). The objective of the study was to examine trends in Objective: contraceptive method choice for young women aged 15-24 years using modern methods and to determine factors associated with their choices. The study used data from KDHS of 2003, 2008/9 and 2014 in Methods: descriptive analysis and logistic regression to determine the socio-economic variables that influence the choice of contraceptive methods for young women. Results showed a general shift in use towards long term modern Results: contraceptives with the shift being more pronounced among young women with primary education, from rural areas, lower wealth households, and low contraceptive use regions. Women with secondary education and higher wealth status are shifting towards short term methods. Findings confirmed socio-demographic factors of age, education, wealth status and type of region as predictors of contraceptive use. Contraceptive information and services should be enhanced Conclusions: for young women to make informed choices concerning their reproductive and sexual health to enable them complete school and transition to colleges to acquire relevant skills that will make them optimally productive and lead Kenya to achieving the demographic dividend.


Introduction
Kenya has invested considerable resources towards improving adolescent sexual reproductive health (ASRH) and has in place the 2015 National Adolescent Sexual Reproductive Health (NASRH) Policy; however, the rates of teenage pregnancies do not reflect these heavy investments and are a matter of public health and social concern. Teenage pregnancy rates have plateaued for over a decade and have persisted at 20% between 2003 and 2014 1-3 .
More than 10,000 Kenyan girls quit school every year due to unintended pregnancies and more than 10 percent of births in Kenya occur to girls aged 15-19 years 4 . Unintended pregnancies among teenagers are linked to poor health outcomes, which include sexually transmitted infections, unsafe abortion, miscarriages, stillbirths and complications during births that can lead to infant and/or maternal deaths. There is also rapid-repeat pregnancy, which is linked to greater risks of morbidity and mortality in mothers and new-borns as well as poor mental health 5 .
The Kenya Demographic and Health Survey (KDHS) 2014 revealed critical facts about adolescents in Kenya. One out of every three (30%) adolescent girls reported having had sex while the average age at which adolescent girls had their first child was 17 years and 12% of the girls were already in a relationship. Most sexually active 15-19-year-olds do not use contraception and only about 50% of those on contraception use it consistently with 14% of them being ambivalent about pregnancy. As many as 90% of unmarried sexually active adolescent girls reported not wanting to get pregnant in the next two years but only 43 percent were using a method to prevent pregnancy. Average unmet need for family planning for the group was 21% against the national average of 18% 1,6 . Several interventions to address ASRH needs have been put in place but the median age at first birth continues to decrease.
Kenya has a very youthful population with the 15-24-yearold age cohort constituting about 30% of the population of 47 million 7 . The 15-24-year-old young women take about 15% which comes to 7 million. With the economic growth far below the desired 10% per annum, this group is critical to economic development but for them to be substantially productive the problems surrounding them have to be comprehensively addressed 8 .
Demographic dividend has been proposed as a viable solution to economic challenges faced by developing countries. Kenya's demographic window is expected to open in 2038 and it could achieve demographic dividend by 2050 if it invests in the huge youth population and especially the adolescents. Demographic dividend modelling shows Kenya can attain $4,595 in GDP per capita by 2050 if it makes strategic investments against $896 if it maintains the status quo. This would mean a staggering fivefold economic growth whose achievement, family planning can accelerate 9,10 . Also, improved access to reproductive health has been identified as critical in the process of development in Sustainable Development Goal (SDG) 3.7 11 . Adolescent and young women childbearing is linked with lower educational achievement and could lead to a cycle of poverty and hinder achievement of demographic dividend. Meeting the contraception needs of adolescent and youth to prevent unintended pregnancies should thus be a priority 12 .
The Health Act of 2017 and constitution of Kenya have enshrined the right of every woman to safe, affordable, effective, and acceptable contraception services 13 . FP2020 has issued the Global Consensus Statement to expand contraceptive options for adolescents and youth and set higher family planning targets 14 . Family planning documents in Kenya advocate for promotion of long acting reversible contraception (implants and IUD) as safe, convenient and highly effective methods.
KDHS data shows the 15-24 age group taking a sizeable share of the modern method mix in Kenya among all users of reproductive age with 15% pill users, 8% of IUD use, 18% implants use and 24% injection use 1 . It also shows shifting popularity from injections to implants for the young women, but correct and consistent use remains a challenge and unintended pregnancies prevail 5 . Data is constantly needed to monitor the success of the many policy and program efforts towards effective reproductive health programs for youth.
Poverty is a major contributing factor to teenage pregnancies with girls from poor households more exposed to the risk of early childbearing 8 . Evidence has shown that without contraception, more than 90% sexually active young women get pregnant within a year 15 . The leading reason for adolescents' delay or for not obtaining contraceptive services is apprehension on confidentiality and side effects.
In Kenya, 22% of unintended pregnancies end up in dangerous abortions for adolescents because they mostly abort during the second trimester after late discoveries of the pregnancies. Adolescents below 19 years make up 20% of women seeking post-abortion care services in health facilities and 50% admissions from severe complications 16,17 . There is urgent need to expand access to safe and affordable reproductive health care and options especially for the more vulnerable young women of lower education and economic status who procure unsafe abortions more 18 .
Evidence has shown increased use of contraception is a primary factor in reduction of adolescent fertility and births rates 19 . Comprehensive sexuality education (CSE) and integrated youth friendly services have also shown success when well implemented but youth centres and peer education have not been effective 20 . Providing contraceptive information, counselling and services and the use Larcs has been found useful to encourage new acceptors of contraception among nulliparous young women. LARCs are 99% effective if correctly used and have several non-contraceptive benefits like reducing menstrual flow and pain and thus endometriosis.
To achieve demographic dividend, a country must undergo a demographic transition which brings about lower birth and death rates through investments in family planning. Significant declines in fertility can be reaped from a high contraceptive use momentum that young women can generate. A slower population growth can also be achieved if age at first birth is delayed for most of these women. A huge young population represents great economic potential and investments in their health and education can stimulate new economic opportunities 21 .
Previous studies in Kenya have identified socio economic factors such as education and wealth as significant on the use of contraception 22,23 . Choice and continued use of a method is influenced by socio-economic factors hence the choice of explanatory variables 24 . This study attempted to provide a current information base on how contraceptive use among 15-24 years old girls in Kenya has evolved from 2003 to 2014 with a view to providing evidence which could form a basis to increase modern contraceptive use among them.
The main objective of the study was to examine factors that influence the choice of contraceptive methods for young women 15 to 24 years old.

Data sources
This study analysed secondary data from the Kenya Demographic and Health Surveys of 2003, 2008/9 and 2014, which contained detailed contraceptive calendars for all interviewed women aged 15 to 49 years. The study was based on the samples of all 15-to-24-year-old women, whether married or single, who reported current use of modern contraception within the five years prior to the three surveys. It excluded those who were not using a main modern contraceptive at the time of the each KDHS. Data was specifically obtained from the variables on current method used, age, education, residence, wealth and region.

Data analysis
Data analysis was performed using SPSS version 22. The study started with obtaining frequencies for each of the methods to obtain a trend analysis of the main methods of contraception used by the young women, namely the Pill, IUD, Implants, Injection and Condoms for the three data sets. Afterwards, young women contraceptive users were profiled using cross-classification analysis on socio demographic variables of age, education, wealth status, place of residence and type of region to provide trends of contraceptive method choice within the period of study. This refers to arranging the data into classes according to the variables. Pearson's Chi square test was used to determine existence of any significant relationships between the dependent variable (contraceptive method choice) and the independent socio demographic variables.
The dependent variable was dichotomous, hence binary logistic regression was used to determine the net effects of the independent variables (age, education, wealth status and type and region of residence) on method choice.
Contraceptive method was recoded as either short term (pills, condoms) or long term (injection, IUD, implants), while level of education was recoded into two categories (primary and secondary/higher). Household wealth status was recoded into three tertiles of lower (lowest, and second lowest), middle (middle) and higher (fourth quintile and highest) while region was recoded into two categories of high contraceptive use (Nairobi, Central and Eastern) and lower contraceptive use (the rest of the regions).

Results
Background characteristics of the study population The first objective of the study was 'to establish the profiles of 15-24-year-old modern contraceptive users' over the years 2003 to 2014. The following Table 1 presents the background characteristics of the study population who reported using a main modern method in each KDHS.
The majority of the study group in 2003 were in the 20-24 age group (76%); this proportion increases to over 80% in 2008/09 and 2014. There is also a trend of rising education levels: in 2003, 68% had received only primary education, which reduces to 51% by 2014; those with secondary education increased from 32% in 2003 to 49% in 2014. The majority The general profile of modern contraceptive users among the group is that of women with primary education, living in urban areas, from higher wealth households and regions of low contraception.

Trends in contraceptive method choice
The section addresses the second objective, 'to establish the trends in contraceptive method choice for young women', 15-24 in Kenya from 2003 to 2014. Further analysis established shares of use for the Pill, Condom, Injection, IUD and Implants presented in the following figure.
As seen in Figure 1, the main method of contraception was injection with a share of 54-58% in the three data sets. Its share increased slightly between 2003 and 2008/09 from 56% to 58% then reduced to 54% in 2014. The pill was second most common in 2003 (23%) but declined to fourth (11%) in 2014.
Condom use increased from 17% in 2003 to the second largest share in 2008/09 at 24% but fell to third position in 2014, below implants. Implants were fourth between 2003 and 2008/09 with a 2-3% share but between 2008/09 and 2014, had a huge upsurge of 16 percentage points from 1.9% to 17.7% to move to second position after injection. The IUD had a very low share of between 0.4% and 2%.

Differentials of contraceptive method choice
The majority of users of long-term contraceptive methods had primary education and lived in rural areas and in regions of lower contraception (Table 2). On wealth status, majority were in the middle category in 2003 but in 2008/09 and 2014, bigger proportions were of lower wealth. Users of short-term methods had secondary education, were wealthy and lived in urban areas mostly in regions of high contraception.

Logistic regression results
Logistic regression results are presented Table 3. They respond to the third objective of the study, 'to identify determinants of contraceptive choice among 15-24 year olds.
Results show that age and education were significantly associated with method choice in 2003. Age was positively associated with choosing a long-term method for a 20-24-yearold woman, being two and a half times more likely than for a 15-19 woman. Education was negatively associated with choosing a long-term method, with women with secondary education being 60% more likely to choose short-term methods than those with primary education.
In 2008/09, age, education and region were significantly associated with choosing a long-term method, with region and age having positive relationships while education continued to exhibit a negative relationship. For age, there was an almost 100%  increase in the odds of a 20-24 woman choosing a long-term method when compared with a 15-19-year-old. Conversely, there was a great decrease in the odds of using a long-term method as the level of education rose. Secondary-educated women were 70% less likely to choose long-term methods relative to those with primary education suggesting an inverse relationship. For region, the odds of a woman from a region of high contraception uptake choosing a long-term method were almost three times greater than for a woman from a region of low contraception uptake.
In 2014, four variables (age, education, wealth status and region) emerged as significant predictors of contraceptive method choice. Age was positively associated while education, region and wealth status were negatively associated. On age, 20-24-year-old women were almost twice as likely to choose a long-term method when compared with their 15-19 cohorts. Education showed a 60% decrease in the odds of a woman with secondary education choosing a long-term method relative to a woman with primary education. Thus women with secondary education were more likely to choose short-term methods than their less educated counter parts.
Region had a negative relationship with method choice and emerged as a very significant explanatory variable influencing method choice. There was a 35% decrease in the odds of a woman from a high contraception region choosing a long-term method when compared with a woman from a region with low contraception uptake. Women from regions of high contraceptive use were thus 65% more likely to choose short-term methods than those from regions of lower contraception uptake. Thus, women from high contraception regions were more inclined to use short-term methods.
Those from higher wealth status were less likely to use long-term methods of contraception than those of lower wealth. Results show the odds of choosing a long-term method decreased with increased household wealth. Young women of higher wealth emerged as 40% less likely to choose a long-term method as compared to those from lower wealth status. Women of higher wealth were 60% more likely to choose short-term methods than those of lower wealth.
There was no notable influence on the choice of a modern method for residence in the regression in all the data sets, suggesting residence could not be used to significantly explain urban-rural variations in contraceptive method choice.
Overall in the regression analysis, age, education, wealth status and type of region emerged as significant predictors of contraceptive method choice.

Discussion
The study has shown use of modern methods among women aged 15-24 years has been increasing from 2003 but more so from 2008/09. Proportions using modern methods increased almost sixfold between 2008 and 2014. Use of modern methods has been increasing globally and in Kenya in recent years for all women of reproductive age, but more so among those aged 15-24 years, where there are huge numbers of sexually active youth who are initiating contraception. The findings are collaborated by the annual FP2020 reports for Kenya 25 . Other studies had reported rising use of modern methods and more so injectables earlier 5,26 .
Use of long-term methods increased by around four times between 2008/09 and 2014. This is not only a Kenyan phenomenon but a general trend in Sub-Saharan Africa, where successful LARC promotion campaigns and innovative youth sexual and reproductive health programs are increasing the numbers of new acceptors. Implants are now the fastest rising method of contraception among women 15-24, their use more than doubling from 10 to 21% between 2008/09 and 2014. Rapid increase in the use of implants has been reported recently for several countries, including Kenya, where scale up campaigns have been conducted 27,28 . LARCs are more effective and cheaper in the long term than shorter-acting contraceptives, and are recommended for most women, including adolescents and nulliparous women 29 .
The shift away from short-term methods resulted in declining use of pills and condoms, as shown in Figure 1 and in latter analysis, as declining use of short-term methods which were previously very popular with young women. Condoms are an important dual contraceptive method for prevention against pregnancy and also sexually transmitted infections and HIV/AIDS. They have been previously marketed aggressively to ensure demand creation but shifting priorities of HIV funding have led to fewer campaigns from 2007, which has probably led to their declining use 30 .
As shown in the trend analysis for individual modern methods, implants are the new contraception frontier for women aged 15-24. Other studies have shown implants as increasing their acceptance among young women as efforts to promote them increase. A study in Kenya to promote LARCs among young women showed a major upsurge of implants between 2008 and 2014 with use doubling in the 15-19 age category and increasing by six times among the 20-24-year-old age group 31 .
Regions of lower contraceptive use are surpassing regions of high contraception and are showing potential for major increases in use of modern contraceptives for women aged 15-24 and it is collaborated by FP2020 progress reports. This presents evidence of potential for scaling up modern CPR and achieving the FP2020 targets 25 .
Region was a socio-economic predictor of contraceptive choice in this study in 2008/09 and 2014. However, the classification of regions in terms of contraceptive use, as performed here, has been changing as seen from KDHS 2014 such that regions which previously had lower contraceptive uptake like Western are recording very high uptake of modern contraception especially for LARCs. This is possibly an outcome of highlevel campaigns to promote long-term methods in the whole country but more so in the regions of lower contraceptive use.
On residence, no distinct variations in method choice are seen in rural or urban areas, suggesting good progress towards equity in accessing methods in both types of residence. Previous studies in Kenya had revealed distinct regional differentials showing the likelihood of using a modern method, more so long-term ones, being higher in urban than rural areas and also higher in regions of high contraception uptake than in regions of lower contraception 22 . In another study in Kenya, residence was also found to influence variations in contraceptive choice 5 .
The findings on the significance of education and wealth status are in line with earlier studies in Kenya that found education and wealth had significant influence on the use of contraception 22,23 . A study in Uganda found similar results 24 . In contrast, other studies on Kenya have not found education and household wealth to be significantly associated to contraceptive use 32,33 .
Results have shown that users of modern contraception methods are exhibiting higher education levels in recent years. This is probably a reflection of the heavy investments in education Kenya is making to ensure all children complete primary school and transit to secondary school, and also introducing free primary and secondary education. However, the results of regression with education showed an inverse relationship between education and choice of long-term contraceptive methods in all the data sets. Women with secondary education were 40-70% less likely to choose long-term methods than those with primary education. The results tally with surprising recent findings in Ghana that women with secondary education were using less effective, short-term and periodic contraception more than effective, long-term, modern methods 33,34 .
Questions arise whether they are experimenting with methods, are wary of side effects or are intending to get pregnant.
An interesting finding is that users of more modern contraception were in the lower wealth brackets in 2008/09 and 2014 and users of long-term methods were more in rural areas. This could imply success of family planning campaigns in rural and low-income areas or improved access for women of lower wealth status. Community health workers (CHWs) whose role in expanding contraceptive access has been revitalized work in some of these rural and hard to reach areas where most households are of lower wealth.
As observed in the study results, 15-24-year-old users of modern methods are shifting away from short-term contraceptives towards LARCs, which are more efficient, convenient and offer greater user satisfaction 35 . The shift is more pronounced among women with primary education, from rural areas, lowerwealth households and regions of low contraceptive use. Implants may overtake the injection as the most popular method for this group if the surge in popularity continues.
What do the results mean for Kenya?
The study results generally indicate success for the aggressive Kenyan family planning program and FP2020 campaigns and existence of great potential for scaling up use of long-term methods among the study group. The phenomenal rise in use of long-term methods observed in the review period will most likely continue due to the ongoing global and local promotion towards LARCs and also because of the momentum of demand from the huge population of 15-24 year olds.
Some program and policy implications arise from the study findings. Young women are shifting towards long-term methods and the use of implants. These are provider-dependent methods and therefore the government needs to invest more in ensuring enough skilled providers to handle insertions/removal as well as counselling on any associated side effects. It is expected that implants, because they are long-term methods, will reduce discontinuations and consequently unintended pregnancies among the young women.
The shift towards long-term methods spells success for the policy on promotion of LARCs for all women, including adolescents. This should increase retention of users of modern methods, protect the gains made in CPR and modern CPR, and contribute towards further reductions in fertility. The cost of LARCs is higher initially, but they are cost-effective in the long term, meaning more funding should be provided towards generating sustainable commodity stocks.
The changing profile of the users of long-term methods, who mostly live in rural and low-income areas, show potential for increased uptake and stepping up the campaigns for LARCs in these areas with more involvement of CHWs.
The emerging shift by secondary educated, wealthier women towards short-term methods calls for targeting of the group to minimise unwanted fertility.
The declining use of condoms raises concern whether the decline will result in increasing incidences of HIV/AIDS and roll back the gains in the fight against HIV/AIDS.

Conclusions
Results from the study have provided a useful profile of young modern contraceptive users and established a general pattern of their contraceptive use. The study has identified the predictors of contraceptive use among young girls as age, education, household wealth and type of region, and detailed how these factors are changing with time. This information is useful for managers of youth reproductive health programs and could guide interventions to help reduce unintended pregnancies among young girls in Kenya so that girls can stay in school longer and complete their education to desired levels.
The results for the young women, 15-24 years of age, conform to the general shift towards long-term methods for all women of reproductive age, as seen in the recent studies 35,36 . Kenya is an example of family planning policy success but the journey towards replacement level fertility cannot be complete if the 15-24 age group is left behind. More investment towards successful adolescent and youth reproductive health are needed to ensure Kenya takes advantage of the opening of the demographic window and reaps the demographic dividend.
Young women aged 15-24 are at the centre of these efforts; family planning policy and program managers must ensure they take control of their reproductive health and use contraceptive information and services to their advantage.
Contraception allows young women to make informed choices concerning their reproductive and sexual health, stay in school longer, complete higher levels of education and transition to colleges for labour relevant skills that will enable them optimally contribute to the economy and lead Kenya to achieving the much desired demographic dividend. LARC has potential advantages for these women who wish to delay pregnancy for several years.
To improve the environment for reproductive health and modern contraception for women 15 to 24 years with a view to minimising unintended pregnancies, several interventions can be put in place: 1) Implement fully the National Adolescent Sexual and Reproductive Health Policy, 2015; 2) Initiate age appropriate Life Skills Education (LSE) in all schools; 3) Mainstream youth friendly services into regular services at all health facilities; 4) Expand the range of available method mix for married and unmarried young women and especially the LARCs; and 5) Ensure commodity security by reducing stock outs of commodities.

Source data
The datasets analysed during the current study are available from the MEASURE DHS repository, (http://www.measuredhs.com). Access to the dataset requires registration and is granted to those that wish to use the data for legitimate research purposes. A guide for how to apply for dataset access is available at: https://dhsprogram.com/data/Access-Instructions.cfm.
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com