Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003-2014.

OBJECTIVES: This study aimed to examine patterns and determinants of modern contraceptive discontinuation among women in Kenya.
METHODS: Secondary analysis was conducted using national representative Kenya Demographic and Health Surveys of 2003, 2008/9, and 2014. These household cross-sectional surveys targeted women of reproductive age from 15 to 49 years who had experienced an episode of modern contraceptive use within five years preceding the surveys from 2003 (n = 2686), 2008/9 (n = 2992), and 2014 (5919). The contraceptive discontinuation rate was defined as the number of episodes discontinued divided by the total number of episodes. Weighted descriptive statistics, multivariable logistic regression analysis, and Cox proportional hazards analysis were used to examine the determinants of contraceptive discontinuation.
RESULTS: The 12-month contraceptive discontinuation rate for all methods declined from 37.5% in 2003 and 36.7% in 2008/9 to 30.5% in 2014. Consistently across the three surveys, intrauterine devices had the lowest 12-month discontinuation rate (6.4% in 2014) followed by implants (8.0%, in 2014). In 2014, higher rates were seen for pills (44.9%) and male condoms (42.9%). The determinants of contraceptive discontinuation among women of reproductive age in the 2003 survey included users of short-term contraception methods, specifically for those who used male condoms (hazard ratio [HR] = 3.30, 95% confidence interval [CI] = 2.13-5.11) and pills (HR = 2.68; 95CI = 1.79-4.00); and younger women aged 15-19 year (HR = 2.07; 95% CI = 1.49-2.87) and 20-24 years (HR = 1.94; 95% CI = 1.61-2.35). The trends in the most common reasons for discontinuation from 2003 to 2014 revealed an increase among those reporting side effects (p = 0.0002) and those wanting a more effective method (p<0.0001). A decrease was noted among those indicating method failure (p<0.0001) and husband disapproval (p<0.0001).
CONCLUSIONS: Family planning programs should focus on improving service quality to strengthen the continuation of contraceptive use among those in need. Women should be informed about potential side effects and reassured on health concerns, including being provided options for method switching. The health system should avail a wider range of contraceptive methods and ensure a constant supply of commodities for women to choose from. Short-term contraceptive method users and younger women may need greater support for continued use.

Introduction

The global focus on family planning program has been on expanding access to modern contraceptives, assuring method choice, overcoming barriers to use, and improving quality of care [1]. Recent global estimates reveal that 63% of women of reproductive age, 15 to 49 years, used some form of contraceptives in 2017, up from 54.8% in 1990 [2,3]. Family planning is a critical component of safe motherhood programs due to its direct and indirect effect on maternal mortality. The contribution of unintended pregnancies on maternal mortality and morbidity is well documented; studies have estimated that effective use of contraceptives could avert up to 44% of maternal deaths [4].

Most public health programs have traditionally targeted non-contraceptive users to understand their reasons for not using them and designed interventions that address the prevailing gaps. However, there has been less programmatic focus on assessing the level of satisfaction among current contraceptive users [5]. While these programs have resulted in an increase in contraceptive use in general, globally, 38% of contraceptive users discontinue use of a method within the first 12 months [6]. Whereas not all contraceptive discontinuation should be of concern, since the fertility desires of women change over time, discontinuation while still in need is a concern because it contributes substantially to the total fertility rate, unintended pregnancies, and induced abortions [7-10]. Furthermore, an analysis of Demographic Health Survey (DHS) data from 34 developing countries revealed that 38% of women estimated with the unmet need of family planning were prior method users who had discontinued use [6]. This underscores the importance of focusing on current users to ensure their fertility needs are met.

Some studies have demonstrated that certain socio-demographic characteristics—such as younger women, higher parity, and unmarried or not in a union—are the most likely determinants of discontinuation [8,9,11], and that discontinuation rates are higher among short-term method users compared to long-acting reversible contraceptive (LARC) users, such as intrauterine devices (IUDs) and implants [8]. For instance, a multicountry DHS analysis revealed that the lowest 12-month discontinuation rate was for IUDs while the highest was for condoms. In the same study, pills, injectables, periodic abstinence, and withdrawal were discontinued by about 40% of users within the first 12 months of use [8].

Analysis of the DHS’s calendar data derived from the women’s questionnaire is the major source of information on contraceptive discontinuation; it contains robust historical data on episodes of contraceptive use, recalled by women month by month, five years preceding the survey [9]. While there are concerns about recall bias and the validity of calendar data based on its complexity and information, an analysis comparing various studies that used the calendar data and other forms of questionnaires established that calendar data performs just as well or better in terms of reliability and validity on capturing information on contraceptive use [12-14]. Moreover, due to lack of a robust longitudinal cohort data that is nationally representative, this calendar data, with its complexities, remains the best source of data for discontinuation analysis.

Several studies have documented that contraceptive discontinuation when in need of a method is a measure of the quality of family planning services because it can be addressed by improved counseling and instituting follow-up mechanisms [15-17]. Understanding factors that affect the discontinuation of modern contraceptive use is crucial to enable family planning programs to identify appropriate strategies to improve the continuous use of modern contraception [5].

A 2007 DHS multicountry analysis on contraceptive discontinuation included data from Kenya’s 2003 DHS [8]. Since then, Kenya has invested in policies to improve the health care and the social environment that promotes increased use of modern contraceptives. In 2007, the country developed and launched its first-ever national reproductive health policy that sought to provide an enabling environment to increase equitable access and improve quality, efficiency, and effectiveness of service delivery at all levels [18]. In 2009, the country further expanded access to contraceptive methods through community-based distribution programs that allowed community health volunteers to provide family planning information and services, such as pills and injectables, to women in hard-to-reach areas [19]. This policy environment, among other interventions, contributed to an increase in the use of modern contraceptives, from 32% in 2003 to 53% in 2014 [20]. In the last two decades, Kenya’s contraceptive method mix (percentage of current modern method users who use the particular method in question), has largely been driven by the uptake of short-term methods though there has been a gradual increase in the use of LARC methods particularly implants. Injectables’ share of the method mix among contraceptive users has fluctuated from 45.6% in 2003, 55.0% in 2008/9, to 47.9% in 2014. Pills has been on a downward trend from 23.3% in 2003, 18.8% in 2008/9, to 14.1% in 2014. Notably, implants’ share of the method mix use has risen from 5.3% in 2003, 4.8% in 2008/9, to 18.2% in 2014. IUD has changed from 7.5% in 2003, 4.0% in 2008/9, and 5.9% in 2014. Condom use has grown from 3.7% in 2003 to 7.9% in 2014 [20-22]. Despite the overall increase in modern contraceptive use over the last 15 years with changes in the method mix, discontinuation still occurs among one-third of contraceptive users in Kenya [20]. There has not been a subsequent analysis of 2008/9 and 2014 data to assess changes in the determinants and reasons for discontinuation with the increase in modern contraceptive prevalence rates and the corresponding shift in the method mix. Thus it is important to examine contraceptive dynamics such as discontinuation so that family planning programs can provide quality services that meets client needs [23]. Furthermore, such data would allow policymakers and program implementers to monitor progress toward achieving international development goals for family planning. This analysis examines the trends and determinants of contraceptive discontinuation using data from 2003, 2008/9, and 2014 Kenya Demographic Health Survey (KDHS). Kenya has not conducted a national demographic health survey since the KDHS 2014 that would provide additional information on changes in the contraceptive use dynamics.

Materials and methods

Secondary data analysis was conducted based on the KDHS 2003, 2008/9, and 2014 datasets, which are publicly available through the DHS program website https://dhsprogram.com/data/available-datasets.cfm. The KDHS is a nationally representative, cross-sectional household surveys that used a two-stage multistage sampling approach, with a sampling frame based on the 1999 and 2009 national census in Kenya. These analyses used data from the women’s questionnaires, which collect data from women of reproductive age (15–49 years) on a range of socio-demographic characteristics and reproductive history. The population included 8195 women (KDHS 2003), 8444 women (KDHS 2008/9), and 31079 women (KDHS 2014). These analyses were restricted to women who reported to have ever used a method of contraception in the five years preceding the survey and had complete contraceptive histories. The final sub-sample included in the analyses were 2686 (KDHS 2003), 2992 (KDHS 2008/9), and 5919 (KDHS 2014). Information on contraceptive use in the DHS is collected in the form of a reproductive calendar. It contains the past monthly history of reproductive events including births, pregnancies, terminations, and episodes of contraceptive use for the five years before the survey. Female respondents were asked about their contraceptive use for each month of the five years prior to the survey. For months in which a woman reported discontinuing the use of a method she was asked the main reason. Excluded from this study were women who had never used contraceptives, women who had indicated that they were pregnant at the time of the survey, infecund women who were self-reported, and women with incomplete contraceptive information.

Variable definitions

Exposure is defined as the duration of use of a specific method within one episode of use. Exposure begins with an initial month of adoption. It would end with self-reported discontinuation or with the month of the interview if the contraceptive method was still being used at the time of the interview.

Contraceptive discontinuation is defined as starting contraceptive use and then stopping for any reason while still at risk of unintended pregnancy [24].

Discontinuation while “in need,” in this paper, this refers to women who are at risk of becoming pregnant, do not want to become pregnant, and are not using contraception [9]. This was operationalized from the reasons given for discontinuation as recorded in the contraceptive calendar data. Reasons for discontinuation including method failure, side effects, health concern, access, cost, wanted more effective method, inconvenient to use, and husband opposed, were considered to be discontinuation while in need, whereas wanting to become pregnant, infrequent sex/husband away, marital dissolution and menopausal were noted not to be in need.

Statistical analysis

Our analysis focused on women who discontinued modern contraceptive methods (pills, IUDs, injectables, implants, and male condoms). Contraceptive methods that were discontinued on the month of the interview or two months prior were censored to avoid bias due to unrecognized pregnancy since many women do not realize they are pregnant in their first trimester [9]. The 2008/9 KDHS had missing data for reasons for contraceptive discontinuation, a variable used in the computation of the determinants, therefore, it was not included in the survival analysis.

To calculate the discontinuation rate and the number of episodes, the KDHS datasets for 2003 and 2014 were first converted into an event file to report episodes. The discontinuation rate was calculated using a life table that generated the net discontinuation rates. The rate of discontinuation was calculated by dividing the number of episodes discontinued in a month by the total number of episodes that reached that duration. This was also calculated for 12 months’ duration for each of the discontinuation reasons. An individual woman may contribute more than one episode to the calculation.

The contraceptive method was tabulated against the reasons for discontinuation and socio-demographic characteristics accounting for sampling weights. Frequencies and percentages were obtained for descriptive statistics. Twelve-month contraceptive discontinuation rates were calculated as shown in Table 2.

Table 2

Twelve-month contraceptive discontinuation rates for specific methods between 2003–2014 in Kenya.

Method	2003	2008/9	2014	Trend P-value
Pills	46.2	44.2	44.9	0.5162
Injectables	31.8	29.6	30.9	0.5471
IUD	12.4	28.4	6.4	0.0502
Implants	15.3	10.2	8.0	0.0204
Male Condoms	59.4	60.2	42.9*	<0.0001
All Methods	37.5	36.7	30.5	<0.0001

A Cox proportional hazard model was used to obtain hazard ratios (HR), 95% confidence interval (95%CI). In the survival analysis, method discontinuation was the dependent variable and the covariates included contraceptive method, age category, residence, education, marital status, religion, number of living children, and wealth quintile. Trend analysis comparing changes in discontinuation rates or proportions between survey years were compared using a Cochrane-Armitage trend test. Results with p-value <0.05 were considered statistically significant. In the trend analysis, the 2003 survey was used as a base and 2014 as the end line. Additional analyses were conducted to explore the profile of contraceptive users and determinants of contraceptive discontinuation among women in need. These results have been presented as supplementary tables for reference only (S1 and S2 Tables).

Results

Characteristics of the study population

The results shown in Table 1 present the profile of women who discontinued use of a modern contraceptive methods. The majority were aged 25–34 years, 54.9% in 2003, 48.8% in 2008, and 53.8% in 2014. There was an increase in discontinuation among women who were Protestant from 68.9% in 2003 to 75.3% in 2014. Most were married or living together with a partner (76.7% in 2003, 75.3% in 2008, and 77.8% in 2014); living in rural areas (72.3% in 2003, 73.2% in 2008, and 51.1% in 2014). A higher percentage had a primary education—60.6% in 2003, 57.9% in 2008, and 52.1% in 2014. Discontinuation in rural areas declined from 72.3% in 2003 to 51.1% in 2014 while it increased among urban residents from 27.7% in 2003 to 48.9% in 2014. There was an increase in the number of women who discontinued in the highest wealth quintile from 30.0% in 2003 to 31.5% in 2014.

Table 1

Profile of women who discontinued use of contraceptives at least three months before the survey.

	Categories	2003		2008/9		2014
		N	%*	n	%*	n	%*
Age categories (Years)	15–19	157	9.8	109	5.7	150	4.6
	20–24	467	29.2	585	30.5	773	23.6
	25–34	879	54.9	939	48.8	1764	53.8
	35–49	99	6.2	289	15.0	595	18.1
Marital status	Never Married	225	14.1	285	14.8	388	11.8
	Married	1228	76.7	1447	75.3	2552	77.8
	Single**	149	9.3	191	9.9	342	10.4
Residence	Urban	444	27.7	515	26.7	1606	48.9
	Rural	1158	72.3	1407	73.2	1676	51.1
Education level	None	70	4.3	76	4.0	99	3.0
	Primary	971	60.6	1113	57.9	1711	52.1
	Secondary	432	27.0	546	28.4	1011	30.8
	Tertiary	129	8.1	188	9.8	461	14.0
Religion	Catholic	419	26.1	417	21.7	636	19.4
	Protestant	1100	68.9	1380	71.8	2472	75.3
	Muslim	60	3.7	92	4.8	112	3.4
	No religion	16	1.0	31	1.6	43	1.3
	Other	7	0.5	2	0.1	19	0.6
Number of living children	None	181	11.3	216	11.2	310	9.4
	1–2	727	45.4	858	44.6	1617	49.3
	3–4	438	27.3	565	29.4	889	27.1
	5+	246	16.1	284	14.8	466	14.2
Fertility intention	Wants another child	805	51.0	936	49.8	1686	52.1
	Undecided	55	3.5	50	2.6	72	2.2
	No more	718	45.5	891	47.4	1480	45.7
Wealth Quintile	Poorest	161	10.0	216	11.3	298	9.1
	Poorer	268	16.8	332	17.3	571	17.4
	Middle	316	19.7	376	19.5	649	19.8
	Richer	376	23.5	419	21.8	730	22.3
	Richest	481	30.0	579	30.1	1032	31.5

*Some figures may not add up to 100% because of rounding.

** Single refers to Divorced/Separated/Widowed.

Contraceptive discontinuation rates at 12 months for specific methods in 2003 and 2014 in Kenya

Table 2 presents the 12-month discontinuation rates. Overall, the 12-month discontinuation rates for all methods decreased from 37.5% in 2003 to 36.7% in 2008 and 30.5% in 2014. The reduction between 2003 and 2014 surveys was statistically significant (trend p-value<0.0001). Condoms had the highest discontinuation rate in 2003 at 59.4%, followed by pills at 46.2%. Similar trends were observed in 2008/9 with highest rates for condoms the 60.2% and pills at 44.2%. In 2014, the survey results showed a slight variation with pills having the highest rate at 44.9%, followed by male condoms at 42.9%. Between 2003 and 2014, the method with the lowest discontinuation rate was IUDs at 12.4% in 2003 and 6.4% in 2014. Implants had the lowest rate in 2008/9 and IUDs at 6.4% in 2014. Between 2003 and 2014 there was a significant decline in discontinuation rates for male condoms (59.4% vs 42.9%, trend p-value<0.0001) and implants (15.3% vs 8.0%, trend p-value = 0.0204), discontinuation rates were not different for pills, injectables, and IUDs.

Survival analysis of determinants of contraceptive discontinuation rates in Kenya, 2003 and 2014

The results of multivariable Cox proportion hazard regression model (Table 3) revealed that in the 2003 survey, the determinants of contraceptive discontinuation included methods of contraceptive used, age of the woman, and marital status. In the 2014 survey, the determinants of discontinuation rates were methods of contraceptive used and the age of the woman.

Table 3

Survival analysis of determinants of contraceptive discontinuation rates in Kenya, 2003 and 2014.

	2003		2014
	Crude HR	Adjusted HR	Crude HR	Adjusted HR
Contraceptive method
Pills	3.43(2.29–5.14)*	2.67(1.79–4.00)*	5.56(4.00–7.72)*	5.25(3.81–7.25)*
Male Condom	4.70(3.03–7.28)*	3.30(2.13–5.11)*	4.88(3.29–7.25)*	3.80(2.58–5.59)*
Injectables	2.09(1.41–3.11)*	1.60(1.07–2.37)*	3.71(2.69–5.12)*	3.32(2.40–4.58)*
Implants	0.48(0.23–0.99)*	0.42(0.20–0.87)*	1.23(0.82–1.83)	1.15(0.77–1.71)
Intrauterine device	Ref	Ref	Ref	Ref
Age category (years)
15–19	3.74(2.99–4.68)*	3.83(2.85–5.14)*	2.33(1.81–3.01)*	2.07(1.49–2.89)*
20–24	2.85(2.41–3.36)*	3.09(2.47–3.87)*	2.13(1.83–2.48)*	1.93(1.58–2.36)*
25–34	1.76(1.50–2.07)*	1.92(1.58–2.33)*	1.54(1.37–1.73)*	1.44(1.25–1.66)*
35–49	Ref	Ref	Ref	Ref
Residence
Urban	0.96(0.86–1.08)	0.99(0.79–1.24)	1.12(1.00–1.25)*	1.08(0.94–1.25)
Rural	Ref	Ref	Ref	Ref
Education
No education	1.36(1.04–1.79)*	1.28(0.94–1.75)	0.97(0.74–1.28)	1.04(0.78–1.39)
Primary	1.29(1.10–1.51)*	1.04(0.85–1.27)	0.86(0.70–1.07)	0.91(0.73–1.13)
Secondary	1.14(0.96–1.37)	1.05(0.85–1.29)	0.90(0.72–1.13)	0.87(0.70–1.09)
Higher	Ref	Ref	Ref	Ref
Marital status
Never Married	1.50(1.27–1.78)*	0.93(0.76–1.13)	1.31(1.05–1.64)*	0.87(0.69–1.08)
Single**|	1.18(1.00–1.39)*	1.20(1.01–1.42)*	1.03(0.88–1.20)	1.08(0.93–1.26)
Married	Ref	Ref	Ref	Ref
Religion
Catholic	0.94(0.63–1.41)	1.00(0.69–1.45)	0.75(0.50–1.13)	0.72(0.47–1.09)
Protestant	0.92(0.62–1.38)	1.03(0.72–1.48)	0.82(0.55–1.21)	0.79(0.52–1.18)
Muslim	0.99(0.63–1.57)	1.12(0.73–1.72)	0.91(0.59–1.42)	0.93(0.59–1.47)
No religion	Ref	Ref	Ref	Ref
Other	0.88(0.47–1.63)	1.00(0.49–2.00)	1.55(0.84–2.86)	1.28(0.69–2.37)
Fertility intention
Wants another child	1.33(1.20–1.47)*	0.95(0.84–1.06)	1.32(1.20–1.46)*	1.00(0.89–1.13)
Undecided	1.25(0.95–1.63)	1.01(0.77–1.31)	0.77(0.58–1.04)	0.75(0.55–1.02)
No more	Ref	Ref	Ref	Ref
Number of living children
1–2	0.72(0.60–0.87)*	1.03(0.82–1.30)	0.66(0.50–0.89)*	0.84(0.63–1.11)
3–4	0.54(0.45–0.66)*	1.01(0.78–1.31)	0.52(0.39–0.69)*	0.77(0.57–1.04)
5+	0.48(0.39–0.58)*	1.24(0.93–1.67)	0.47(0.35–0.64)*	0.85(0.61–1.18)
None	Ref	Ref	Ref	Ref
Wealth quintile
Poorest	1.18(0.98–1.42)	1.22(0.92–1.62)	0.90(0.76–1.06)	1.04(0.83–1.29)
Poorer	1.14(0.97–1.34)	1.16(0.89–1.52)	0.91(0.78–1.06)	1.01(0.83–1.23)
Middle	1.04(0.90–1.20)	1.07(0.84–1.37)	0.95(0.82–1.11)	1.07(0.88–1.31)
Richer	0.96(0.83–1.11)	0.99(0.79–1.25)	0.84(0.72–0.98)*	0.94(0.79–1.12)
Richest	Ref	Ref	Ref	Ref

* means p-value<0.05

** Single refers to Divorced/Separated/Widowed.

In the 2003 survey, compared to those who discontinued use of IUD, discontinuation was more likely among those who used pills (adjusted hazard ratio [aHR] = 2.67; 95% CI = 1.79–4.00), male condom (adjusted hazard ratio [aHR] = 3.30; 95% CI = 2.13–5.11) and injectables (aHR = 1.60; 95% CI = 1.07–2.37), but implant users were less likely to discontinue use (aHR = 0.42; 95% CI = 0.20–0.87). Similarly, in the 2014 survey, compared to those who discontinued IUD, users of pills (aHR = 5.25; 95% CI = 3.81–7.25), male condom (adjusted hazard ratio [aHR] = 3.80; 95% CI = 2.58–5.59) and injectables (aHR = 3.32; 95% CI = 2.40–4.58) had a higher hazard ratio of discontinuation.

In the 2003 survey, an increase in age was associated with a decreased hazard ratio of discontinuation. Women aged 15–19 years were more likely to discontinue use compared to those aged 35–49 years (aHR = 3.83, 95% CI = 2.85–5.14), and women aged 20–24 (aHR = 3.09, 95% CI = 2.47–3.87). The same pattern is observed in 2014 where an increase in age was associated decreasing hazard rates of discontinuation.

In the 2003 survey, separated/widowed women had a higher hazard ratio of discontinuation compared to women who were married (aHR = 1.20; 95% CI = 1.01–1.42). However, in the 2014 survey, there were no significant association between marital status and discontinuation (Table 3).

Reasons for discontinuation in Kenya, 2003 to 2014

Table 4 shows the overall trend in the main reasons reported for discontinuation of all episodes among women in need of contraceptive between 2003 and 2014. In 2003, a total of 2297 women reported 3041 episodes while in 2014, 5029 women reported 6961 episodes.

Table 4

Reasons for discontinuation while still in need.

	2003			2014
Reasons	% of episodes*	No. of women	% of women	% of episodes*	No. of women	% of women	Episode trend test
Method failure	15.7	336	14.6	10.8	501	10	<0.0001
Husband disapproved	3.5	79	3.5	1.4	56	1.1	<0.0001
Side effects	25.4	608	26.5	29.0	1385	27.5	0.0002
Health concerns	3.2	65	2.8	na**	na	na	na
Access/availability	1.8	45	1.9	0.8	44	0.9	<0.0001
Wanted a more effective method	3.9	63	2.7	8.7	442	8.8	<0.0001
Inconvenient to use	3.4	70	3	1.9	107	2.1	<0.0001
Up to God/fatalistic	na	na	na	0.1	6	0.2	na
Cost too much	1	23	1	0.8	46	0.9	0.3193
Total episodes	3041	2297		6961	5029

*Percentages do not sum to 100% since women who discontinued while not in need of contraceptives, including those who reported they wanted to become pregnant/experienced menopause (2003, 42.1% and in 2014, 46.5%) were excluded from the analysis.

na* implies the reason for discontinuation was not collected in the questionnaire for that particular year.

From the analysis, side effects remain the primary reason for discontinuation in the two surveys, with a significant increase from 25.4% in 2003 (reported by 608 women) and 29.0% in 2014 (reported by 1385 women), observed between the two surveys (p = 0.0002). There was a significant reduction in women reporting becoming pregnant while using contraceptives due to method failure, 15.7% in 2003 and 10.8% in 2014; (p<0.0001) and among those reporting husband disapproval, 3.5% in 2003 and 1.4% in 2014; (p<0.0001). The analysis further revealed an increase in discontinuation for women who wanted a more effective method, 3.9% in 2003 and 8.7% in 2014; (p<0.0001).

Trend analysis of method discontinuation rates by reasons in Kenya 2003 and 2014

Table 5 presents method discontinuation rates broken down by reasons for discontinuation. Among injectable users, there was a significant increase in the proportion of discontinuation due to a desire to switch to another method from 2003 to 2014 (7.1% in 2003 to 9.9% in 2014, trend p = 0.0027) and need for a more effective method (0.3% in 2003 to 1.9% in 2014, trend p = 0.0001). Even though pills and injectables contributed largely to discontinuation due to side effects, there was a significant reduction between the two surveys; discontinuation rates of pills due to side effects declined from 23.3% in 2003 to 15.7% in 2014 (p<0.0001) and injectables from 18.8% in 2003 to 14.3% in 2014 (p<0.0002). Similarly, a significant decrease in the level of discontinuation due to method switching was observed among condom users (8.8% in 2003 to 4.6% in 2014, p = 0.0031) users.

Table 5

Difference in method discontinuation rate by reason among women in need of contraception in Kenya between 2003 and 2014.

	Method failure			Side effects/health concerns			Switch to another method			Wanted a more effective method			Other method related reasons*
Methods	2003	2014	P-value	2003	2014	P-value	2003	2014	P-value	2003	2014	P-value	2003	2014	P-value
Pills	4.0	5.3	0.1318	23.3	15.7	<0.0001	12.4	19.8	<0.0001	2.4	6.8	<0.0001	4.7	2.2	0.0003
Injectables	1.0	1.7	0.0777	18.8	14.4	0.0002	7.1	9.9	0.0027	0.3	1.9	0.0001	2.6	1.2	0.0004
IUD	0.6	0.8	0.8373	7.9	4.2	0.1393	1.9	3.8	0.3469	0	0	NA	0	0.2	0.645
Implants	0.7	0.3	0.5351	12.0	6.6	0.0604	1.2	3.4	0.2723	0	0.1	0.7718	0	0.1	0.7718
Condoms	3.7	1.9	0.0536	0.3	0.8	0.3313	8.8	4.6	0.0031	3.8	2.3	0.1321	10.1	0.3	<0.001

*Includes: access/availability, inconvenient to use, and cost. P-values generated from Cochran trend test; IUD, intrauterine device.

Discussion

Our analysis has established that the contraceptive discontinuation rate in Kenya among modern method users significantly declined from 37.5% in 2003 to 30.5% in 2014. This rate is slightly lower than those observed in other sub-Saharan Africa countries, including Ghana (54.0%), Ethiopia (37.1%) Tanzania (37.7%), and Malawi (34.1%) [8,25,26].

The decline in contraceptive discontinuation rate observed between 2003 and 2014 may be attributed to the increase in the modern contraceptive prevalence rate—53% in 2014, up from 32% in 2003—with a corresponding increase in LARC uptake—13.3% in 2014 compared to 4.1% in 2003 [20,22]. Following the 2012 London Summit on Family Planning, the increased focus on family planning, globally and in Kenya, could explain the decrease in contraceptive discontinuation as attention was placed on scaling up of new methods, which expanded the method mix, and building the capacity of health care workers on the provision of contraceptive methods, particularly implants, and engaging the community to support family planning [27].

Results of the survival analysis established that the contraceptive methods used and the age of the woman were the main determinants of discontinuation. The 2003 survey also identified marital status (single) as an additional determinant. Discontinuation was more likely to be reported among users of short-term methods of contraceptives and among younger women. Additional exploration of the data to understand whether these determinants were different among women in need of contraceptive showed that the results were similar (supplementary file, S1 Table). These findings are consistent with studies done in Ethiopia and Senegal and in the DHS analysis of 60 countries in Africa, Asia, Eastern Europe, and Latin America that indicated short-term methods have the highest discontinuation rates [8,9,15,28]. In Kenya, use of short-term contraceptive methods—specifically injectables, pills, and condoms—among women of reproductive age, increased from 23.0% in 2003 to 36.4% in 2014 [20,22]. Removal of LARC methods such as IUDs and implants, requires a health care worker, which could explain the lower discontinuation rates, as compared to short-term contraceptive methods such as condoms, pills, and injectable contraceptives, which can be abandoned by the users without interaction with health care workers [9,29,30].

Our data presented as supplementary file, S2 Table indicated that LARC users were more likely to be older women and were more likely to be using the methods for limiting their family size, thus less likely to discontinue. Use of permanent contraceptive methods, particularly female sterilization was quite low at 3.2% in 2014 [20], hence more women who no longer desire to have children were more likely to use a LARC method, which confers longer-term protection as opposed to a short-term method that require frequent resupply.

The trend analysis revealed that side effects continue to be the leading reason for discontinuation among women, followed by method failure. These findings are corroborated in other studies conducted globally [7-9,11,26,31]. When coupled with results that indicate short-term method users are more likely to discontinue, we posit that discontinuation due to side effects occurred mostly among users of injectables and pills. This is corroborated by a study conducted in sub-Saharan Africa that reported that users of pills and injectables who experienced side effects that were not tolerable, mostly due to changes in bleeding patterns, discontinued use or switched to a method perceived to be more tolerable [32]. No modern contraceptive method is free of side effects, however, epidemiological studies conducted over the last four decades, largely in the United States and Europe, to evaluate the health effects associated with the use of these methods have established that the benefits of contraceptive use outweigh health risks [28]. Several studies have also indicated that the fear of side effects/health concerns whether perceived or real can influence a woman’s decision to discontinue use of a contraceptive method, or deter potential new users [9,15,17,28]. These concerns need to be taken seriously and addressed.

Other notable reasons for contraceptive discontinuation were switching to another method or wanting a more effective method; this was especially true for women who used pills, injectables, and condoms. Women who use short-term method must make a conscious effort to maintain consistent use of their contraceptive and are therefore more prone to discontinuation [33]. Implants were not widely available in Kenya during the 2003 and 2008/9 surveys, which probably explains their low uptake of 1.7% and 1.9% respectively [21,22]. It is worth noting that between 2003 and 2014, implants contraceptive prevalence rate increased from 1.7% to 9.9%, which may explain the shift from other methods to implants [20]. Access to implants tremendously increased in Kenya, and other developing countries, following the 2012 London Summit, which resulted in commitments by countries, donors, and pharmaceutical companies to increase access to contraceptives, including implants, reduce the cost of commodities, and guarantee supplies of contraceptives [27]. Kenya implemented an implant scale-up program that saw the nationwide roll-out of the method as part of expanding the method mix, which led to a significant shift, mostly observed among short-term method users, underscoring the desire by women to get a method that confers long-term protection. Additionally, studies have shown that, during the first year of typical use, LARC methods are more effective than short-term methods: implants are 120 times more effective than injectables, and 180 times more effective than pills; 6% of women using an injectable and 9% of women using pills experienced an unintended pregnancy during the first year of use compared to 0.05% of women with an implant and 0.8% of women with an IUD [34,35]. These realities could explain the reason why women want a more effective method. However, to get users’ perspectives, additional studies need to be undertaken to understand women’s desire to have more effective methods.

National family planning programs should intentionally put in place strategies to address contraceptive discontinuation. Side effects are a major concern among current and potential contraceptive users. As part of provision of quality family planning services, during the initiation of contraceptive use, women ought to be provided with information on potential side effects and options to consider, including method switching, when they experience intolerable side effects. Health education should be provided to women to allay potential fear of side effects.

Technological advancement is needed to support manufacturing of contraceptive technologies that are better tolerated by women. Expanding available contraceptive options ensures that women have a wide variety of methods to choose from if they are not satisfied with their current method.

The package of services offered by community health volunteers at the household-level should include follow-up with contraceptive users, especially short-term method users, to establish their level of satisfaction with their current method, to support users to ensure consistent use while in need, and to link women to health facilities when they experience concerns about the method to improve continuation. With the high mobile phone penetration of over 80% in Kenya, use of mhealth should also be embraced as a channel for provision of contraceptive information to women, including sending reminders to short-term method users to avoid an unintentional discontinuation [36].

One of the major strengths of our study was the use of large, nationally representative survey data, which makes the findings generalizable and applicable to countrywide policies and interventions. The multi-year analysis included data collected for 15 years; this allowed us to monitor changes in contraceptive use dynamics and the impact of policies and programs rolled out over the same period. It also provides an opportunity for future analysis. Nonetheless, there are certain limitations to our study. Although the findings were interesting and insightful, DHS data are cross-sectional, hence causal relationships cannot be established. DHS also do not collect data on the quality of family planning services that would allow us to assess the relationship between discontinuation and quality of care received. Our analysis used the contraceptive calendar data that is collected for five-year period prior to the survey. Women were asked to recall their contraceptive use history, month-by-month basis, which could result to recall bias. Our findings should be interpreted in the context of these limitations. However, the significance of findings can be improved with a future analysis when less biased data becomes available. Another limitation is that data collected only one main reason for discontinuation, while women might have multiple reasons for discontinuation.

Conclusion

Our study reports wide variation in contraceptive discontinuation rates by method, with lower probabilities of discontinuation of highly effective methods, such as the long-acting reversible methods. The study also demonstrates a significant reduction in contraceptive discontinuation between 2003 and 2014. With the increase in the use of modern contraceptives, programs should monitor trends in contraceptive use dynamics, including discontinuation and method switching. Public health programs should strengthen service quality, including the supply chain system, and enhance the provision of information that is client-centered as part of a rights-based approach for family planning services. Despite the improvement in service quality, some women may still choose to discontinue use of a contraceptive while in need of a method, which underscores the need to expand available options.

Longitudinal studies to assess women’s contraceptive dynamics, including discontinuation and method switching, can better inform programs and address contraceptive discontinuation. More studies should also be conducted to understand the level of information on side effects that women are provided by health care providers during initiation of contraceptive use and the impact the information might have on contraceptive continuation. In addition, further studies are needed to understand whether the side effects that women report in the DHS, reflects their lived experience of FP use or their fears and perception since DHS does not distinguish between the two. Understanding and responding to why women discontinue use of a method while still in need should increase use of modern contraceptives, hence reduce unintended pregnancies and, in turn, maternal morbidity and mortality, improving the lives of women and their families.

Survival analysis of determinants of contraceptive discontinuation while in need in Kenya 2003 and 2014.

(PDF)

Click here for additional data file.

Profile of ever users of family planning methods.

(PDF)

Click here for additional data file.

14 Aug 2020

PONE-D-20-19927

Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003–2014

PLOS ONE

Dear Dr. Ontiri,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 28 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Catherine S. Todd

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Additional Editor Comments (if provided):

In addition to the points raised by the reviewers, I recommend making both the Introduction and Discussion more concise and reading through to edit areas with awkward phrasing (such as over-use of the word "majority" in the first part of the Results or using colloquial expressions like "backed-up"). Please clarify how "in need" was defined for the purposes of this analysis when determining which women were retained. While the focus of the paper is to report discontinuation, it is difficult to put these rates into perspective without also having data for method prevalence/use so please add this to the Results. Last, I would like to see greater nuance in contextualizing discontinuation rates. Were women who used the IUCD older overall and thus using the IUCD to limit rather than space? These women may have different thresholds or reasons for discontinuation as compared to women who are using COCs provided at the community level. For the expanded community-level provision of some short-acting methods, how likely is supply chain to factor into continuation vs. actual rejection of the method? Overall, this is a well-written manuscript and if the issues raised here and with the reviewers are addressed, should soon be acceptable for publication.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is a nice paper that utilized the complex DHS contraceptive calendar data to look at trends in contraceptive discontinuation rates in Kenya over three surveys. While I think the overall analysis and interpretation are sound, there are several key areas for the authors to improve, outlined below. The main issues are

1) the DHS dataset is complex, while also having potential concerns about validity given the reliance on 5 year recall, month-by-month for women. The authors need to explain this in a little more detail to both help the reader understand the complexity and to address head-on concerns about the validity of the interpretation.

2) the authors need to describe the statistical test and results they did to compare the trend of contraceptive discontinuation rates across the surveys.

Line 95: Change the sentence “Kenya’s 2003 DHS of 2003” to not repeat the year.

Overall, I think the introduction did a good job of providing important context to this study, however it becomes repetitive in the third paragraph (lines 65-72 state the same concept—contraceptive discontinuation while not wanting pregnancy—in multiple ways), and does not describe the strengths and weaknesses of the contraceptive calendar to allay readers’ potential concerns about the strength of the data.

Materials and Methods: Please add a sentence about how contraceptive calendar data is collected/organized (recall of family planning need and contraceptive use for each month for the previous 5 years).

Line 145/6: The sentence. “the datasets for 2003 and 2014 were first converted into an event file because the data was calendar data” is not clear.

Results:

Line 176-177: Could the authors calculate the statistical significance of the decrease in contraceptive discontinuation across the three time periods? This is also stated as the third sentence in the discussion (“no significant change”, but no discussion of whether a statistical test was used).

Table 2 would be improved with actual numbers, so we can see what proportion of the population was using each method.

Table 3: The ** and *** are not defined on the table.

Table 4: The authors need to describe the decision to classify “it’s up to god, or fatalism” as an effect modifier and the rationale for excluding it from the analysis.

Table 5 would be more accurately titled something like “Difference in method method continuation rates by reason, among women in need of contraception in Kenya between 2003 and 2014.”

Discussion:

Line 240: “A declined in contraceptive discontinuation between 2008 and 2014 was observed” seems to belie the preceding statement, lines 237-238, stating that no significant change was observed between 2003 and 2014—there are no statistical tests shown that support those statements.

Reviewer #2: Comments on PONE-D-20-19927

This paper addresses an important topic. It is well-written, nicely framed, and uses appropriate analytical methods. It was a pleasure to read. I have two primary comments that I raise for the authors’ consideration in their treatment of the discussion/conclusion and abstract, followed by several very minor comments.

Recommendations to prioritize short-term method users and young women seem to come from the finding that these users discontinue at higher rates than users of other methods/older women. However, discontinuation is not necessarily bad or a signal that something in wrong. Just discontinue while still in need (as the authors themselves point out in the introduction). There may be a selection effect: young women choosing condoms because it is easy to discontinue these methods when needs change. Focusing on these women/long-term methods writ large could actually risk ignoring their needs and impairing informed choice. This is more nuanced than is expressed in the abstract (the discussion does a slightly better job of this). I like the focus on resupply of short-term methods.

Similarly, the recommendation to focus on counseling because of the increasing share of reasons for discontinuation being side effects. 1. Did the authors analyze whether the relative increase in side effects as a reason is accompanied or not by an absolute increase in this reason? I.e. are more women discontinuing due to side effects? (Same Q for wanting a more effective method). I’m wary of responding to concerns of side effects with better counseling about side effects. There is a large literature (more developed in high-income countries, but existing globally) about the medical establishment discounting women’s reports of pain and other symptoms. There is also a literature about women’s experiences with side effects. The DHS measure “side effects/health concerns” cannot distinguish whether it is women’s experiences or fears of side effects. We should be careful not to assume that it’s all fears in women’s heads. How do the recommendations to enhance counseling, particularly on side effects, respond to women’s needs if it is women’s experiences of side effects prompting discontinuation? What other recommendations might we consider to ensure that women who experience side effects with their selected method nonetheless have their contraceptive needs met? The issues of counselling and side effects should be a little more developed in the discussion and conclusions section.

Minor comments:

Data are available through The DHS Program website. MEASURE DHS was an old contract and references to that name should be updated. Likewise, ICF International is now ICF.

Proofread: In at least one instance, the 2008-09 Kenya DHS appears as “2008/8” (p2, line 27).

Do you want more consistency between the terms “family planning” and “contraception” (or modern contraception) or is this distinction intentional?

Is uptake distinguishable from use? I believe, when discussing prevalence, use may be the more accurate term (p5, line 89).

P6, line 122: Do you mean “adoption” instead of “application”?

P7, line 126: It would more consistent to use “episodes” in lieu of “incidents”.

Can Table 1 be formatted to have some separation between or better alignment between the N and (%)? It is a little difficult to read as is.

When discussing changes in 12-month discontinuation rates across surveys (p10), it may be worth noting any substantial shifts in the method mix over that time period.

Please use aHR consistently. On p11, lines 199-200 refer to the aHR, but have it labeled HR, making it unclear whether it is the crude or adjusted hazard rate, without referring to the table. (Alternately, the crude HRs could be removed from Table 3.)

In Table 4 (p13), the na* on the row for “up to God/fatalistic” is in bold instead of plain text.

Presentation of data on reasons for discontinuation (p13) exclude all fertility/need-based reasons. I assume that these reasons make up the difference between the sum of the reasons presented in Table 4 and 100%, i.e. that the authors are presenting the relative share of each reason to all reasons for discontinuation. This should be clarified, particularly if I am wrong in this assumption.

In the discussion (p16), the authors summarize their results in terms of likelihoods. While there is a correspondence between likelihoods and hazards or hazard ratios, technically, the authors calculated the latter, not the former. I’m not sure this is a point worth getting worked up about, though.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Dr. Kerry L.D. MacQuarrie

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Comments on PONE-D-20-19927.docx

Click here for additional data file.

25 Sep 2020

Re: Revision of Manuscript “Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003–2014”

This paper addresses an important topic. It is well-written, nicely framed, and uses appropriate analytical methods. It was a pleasure to read. I have two primary comments that I raise for the authors’ consideration in their treatment of the discussion/conclusion and abstract, followed by several very minor comments.

1. Recommendations to prioritize short-term method users and young women seem to come from the finding that these users discontinue at higher rates than users of other methods/older women. However, discontinuation is not necessarily bad or a signal that something in wrong. Just discontinue while still in need (as the authors themselves point out in the introduction). There may be a selection effect: young women choosing condoms because it is easy to discontinue these methods when needs change. Focusing on these women/long-term methods writs large could actually risk ignoring their needs and impairing informed choice. This is more nuanced than is expressed in the abstract (the discussion does a slightly better job of this). I like the focus on resupply of short-term methods.

Response: We take cognizant of the issue raised. We conducted further exploratory analysis and established that the determinants of discontinuation among women who need contraceptive are still the same (short-term method users and younger women). These results are provided as supplementary files and not included in the main paper. Use of short-term methods was associated with higher rates of discontinuation.

We have revised our recommendation as indicated in this excerpt from the abstracts as follows- “We recommend that Family planning programs should focus on improving service quality to strengthen the continuation of contraceptive use among those in need. Women should be informed about potential side effects and reassured on health concerns, including being provided options for method switching. The health system should avail a wider range of contraceptive methods and ensure a constant supply of commodities for women to choose from. Short-term contraceptive method users and younger women may need greater support for continued use.”

2. Similarly, the recommendation to focus on counseling because of the increasing share of reasons for discontinuation being side effects. 1. Did the authors analyze whether the relative increase in side effects as a reason is accompanied or not by an absolute increase in this reason? I.e. are more women discontinuing due to side effects? (Same Q for wanting a more effective method).

Response: Based on this comment, we have added a column in Table 4 to indicate the number of women contributing to these episodes and proportion of women. The increased share of reasons for discontinuation is accompanied by increased number of episodes (and proportion of women reporting those episodes.

3. I’m wary of responding to concerns of side effects with better counseling about side effects. There is a large literature (more developed in high-income countries, but existing globally) about the medical establishment discounting women’s reports of pain and other symptoms. There is also a literature about women’s experiences with side effects. The DHS measure “side effects/health concerns” cannot distinguish whether it is women’s experiences or fears of side effects. We should be careful not to assume that it’s all fears in women’s heads. How do the recommendations to enhance counseling, particularly on side effects, respond to women’s needs if it is women’s experiences of side effects prompting discontinuation? What other recommendations might we consider to ensure that women who experience side effects with their selected method nonetheless have their contraceptive needs met? The issues of counselling and side effects should be a little more developed in the discussion and conclusions section.

Response: This is well noted. We have now qualified the recommendation on counseling to include counseling on other available methods to encourage women to switch when and if they experience side effects or health concerns. In the discussion section, we have alluded to the fact that women are not given adequate information on the potential side effects and what to do when they experience them and provided a case for method switching where possible. Furthermore, we have also added a recommendation on investing more in contraceptive technology to provide and array of options that are better tolerated by women. We also do appreciate that some of the reported side effects may be based on fears. On this, we have now recommended information to be provided to women to allay their fears which include dispelling myths and misconceptions to ensure usage while in need. And furthermore, we have recommended additional studies to explore whether the side effects that women cite is based on their experience or their fears.

Minor comments:

4. Data are available through The DHS Program website. MEASURE DHS was an old contract and references to that name should be updated. Likewise, ICF International is now ICF.

Response: This is noted and have been revised accordingly as follows: The datasets used in the analysis are publicly available and can be accessed online on The DHS Program website https://dhsprogram.com/datasets

5. Proofread: In at least one instance, the 2008-09 Kenya DHS appears as “2008/8” (p2, line 27).

Response: This has been changed in line 28

6. Do you want more consistency between the terms “family planning” and “contraception” (or modern contraception) or is this distinction intentional?

Response: We have revised the write up to ensure consistency between the terms family planning and modern contraceptives. We have now dropped contraception, and adopted use of family planning when referring to the services that allow individuals to achieve desired birth spacing and family size, and timing of pregnancy, and contraceptive (or modern contraceptives) when referring to the actual methods such as IUD, implants that individuals use.

7. Is uptake distinguishable from use? I believe, when discussing prevalence, use may be the more accurate term (p5, line 89).

Response: That has been revised, and we have adopted the term use (line 100)

8. P6, line 122: Do you mean “adoption” instead of “application”?

Response: This has been changed to adoption (line 146)

9. P7, line 126: It would more consistent to use “episodes” in lieu of “incidents”.-

Response: This section where this comment was made (definition of contraceptive discontinuation rate) has been revised after we noticed repetition of this definition in statistical analysis section. In the statistical analysis section, lines 161-167 we have used “episodes” as opposed to “incidents” for uniformity.

10. Can Table 1 be formatted to have some separation between or better alignment between the N and (%)? It is a little difficult to read as is.

Response: This has been revised.

11. When discussing changes in 12-month discontinuation rates across surveys (p10), it may be worth noting any substantial shifts in the method mix over that time period.

Response: This analysis was done by DHS, and we did not repeat it in our results since our paper is focusing on discontinuation. However, we have included a paragraph in our introduction section on to indicate the shifts in the method mix (lines 100-112. This has further been referenced in our discussion (lines 286-287, 295, 317-319

12. Please use aHR consistently. On p11, lines 199-200 refer to the aHR, but have it labeled HR, making it unclear whether it is the crude or adjusted hazard rate, without referring to the table. (Alternately, the crude HRs could be removed from Table 3.)-

Response: This has been revised accordingly in lines 222-223

13. In Table 4 (p13), the na* on the row for “up to God/fatalistic” is in bold instead of plain text.

Response: This has been revised

14. Presentation of data on reasons for discontinuation (p13) exclude all fertility/need-based reasons. I assume that these reasons make up the difference between the sum of the reasons presented in Table 4 and 100%, i.e. that the authors are presenting the relative share of each reason to all reasons for discontinuation. This should be clarified, particularly if I am wrong in this assumption.-

Response: The percentages don’t add up to 100% since women who were not in need of contraceptives (2003, 42.1% and in 2014, 46.5%) were excluded from the analysis. These included women who reported they wanted to become pregnant/menopause.

15. In the discussion (p16), the authors summarize their results in terms of likelihoods. While there is a correspondence between likelihoods and hazards or hazard ratios, technically, the authors calculated the latter, not the former. I’m not sure this is a point worth getting worked up about, though.

Response: The language has been revised to report hazard ratios for consistency.

16. Additional Editor Comments (if provided):

In addition to the points raised by the reviewers, I recommend making both the Introduction and Discussion more concise and reading through to edit areas with awkward phrasing (such as over-use of the word "majority" in the first part of the Results or using colloquial expressions like "backed-up").

Response: We have revised the introduction and discussion section to be more concise and removed the awkward phrases.

17. Please clarify how "in need" was defined for the purposes of this analysis when determining which women were retained.

Response: This has been defined in the variable section as women who are at risk of becoming pregnant, do not want to become pregnant, and are not using contraception. A reference to this definition has been provided (lines 151-152)

18. While the focus of the paper is to report discontinuation, it is difficult to put these rates into perspective without also having data for method prevalence/use so please add this to the Results.

Response: As mentioned in our response to comment 11, this analysis was done by DHS, and we didn’t not repeat it in our results since our paper was focusing on discontinuation. However, we have included a paragraph in our introduction on the contraceptive prevalence use and method mix, which has also been referenced in our discussion to assist during the interpretation of the discontinuation rate

19. Last, I would like to see greater nuance in contextualizing discontinuation rates. Were women who used the IUCD older overall and thus using the IUCD to limit rather than space? These women may have different thresholds or reasons for discontinuation as compared to women who are using COCs provided at the community level. For the expanded community-level provision of some short-acting methods, how likely is supply chain to factor into continuation vs. actual rejection of the method? Overall, this is a well-written manuscript and if the issues raised here and with the reviewers are addressed, should soon be acceptable for publication.

Response: We have further conducted additional analysis to show the socio-demographic profile of women who used various methods in 2003 and 2014 DHS, which has been presented as supplementary file 1. This analysis has established that most of the LARC users (implants and IUDs) were likely to be older women, hence most likely using the method for limiting. We have further included this aspect in our discussion section, lines 292-297.

The comment on the expanded community-based distribution is valid. It is true that supply chain issues especially on availability of the commodities has an impact on continuation or rejection of a method; if women feel they might be using a method that doesn’t have consistent supply they might opt out. However, this reason did not feature in our analysis. In our recommendation, we have made a case to strengthen service quality which includes the supply chain system both at facility and community level.

20. Reviewer #1: This is a nice paper that utilized the complex DHS contraceptive calendar data to look at trends in contraceptive discontinuation rates in Kenya over three surveys. While I think the overall analysis and interpretation are sound, there are several key areas for the authors to improve, outlined below. The main issues are

1) the DHS dataset is complex, while also having potential concerns about validity given the reliance on 5 year recall, month-by-month for women. The authors need to explain this in a little more detail to both help the reader understand the complexity and to address head-on concerns about the validity of the interpretation.

Response: Additional information on the potential bias and validity concern on the use of DHS calendar data has been provided in the introduction section lines 75-84; and the limitation in lines 363-369.

21. 2) the authors need to describe the statistical test and results they did to compare the trend of contraceptive discontinuation rates across the surveys.

Response: In line 177 we have stated that we used Cochrane-Armitage trend test to compare discontinuation rates across the survey years.

22. Line 95: Change the sentence “Kenya’s 2003 DHS of 2003” to not repeat the year.

Response: This change has been made in line 92

23. Overall, I think the introduction did a good job of providing important context to this study, however it becomes repetitive in the third paragraph (lines 65-72 state the same concept—contraceptive discontinuation while not wanting pregnancy—in multiple ways), and does not describe the strengths and weaknesses of the contraceptive calendar to allay readers’ potential concerns about the strength of the data.

Response: The introduction section has been revised to remove the repetition. Additional details on the strengths and weaknesses of the contraceptive calendar has been provided in the introduction section, lines 75-84; an analysis comparing various studies that have used the calendar data and other forms of questionnaires, have established that the calendar data performs just as well or better in terms of reliability and validity on capturing information on contraceptive use

24. Materials and Methods: Please add a sentence about how contraceptive calendar data is collected/organized (recall of family planning need and contraceptive use for each month for the previous 5 years).

Response: This has been added from lines 77-78 and 364-366

25. Line 145/6: The sentence. “the datasets for 2003 and 2014 were first converted into an event file because the data was calendar data” is not clear.

Response: This has been revised to indicate that the calendar data was converted into an event to enable calculation of discontinuation rates (lines 161-163)

26. Results:

Line 176-177: Could the authors calculate the statistical significance of the decrease in contraceptive discontinuation across the three time periods? This is also stated as the third sentence in the discussion (“no significant change”, but no discussion of whether a statistical test was used).

Response: Table 2 has been revised to include statistical tests comparing the average trend from 2003, 2008 and 2014. The trend test assumes linearity of change from 2003 to 2014, hence reported as net change. We have further corrected the term ‘no significant change’ to ‘significant decrease’ after the statistical test indicate p-value <0.0001

27. Table 2 would be improved with actual numbers, so we can see what proportion of the population was using each method. We have included the actual number of episodes for each method.

Response: We agree that Table 2 would have been improved if we reported the number of episodes or women. However, we note the complexity of reporting this. First, the calculation of 12-month discontinuation for each woman differs because each woman had different start and end dates for a 12-month duration for each episode. Secondly, some women discontinue multiple methods within the 12-month of the start of each episode and accounting for multiple discontinuations from multiple switchers, become more complex. We, therefore, keep the discontinuation rates without the proportion of the population of women

28. Table 3: The ** and *** are not defined on the table.

Response: This has been revised and defined. The asterisk * to refer to significant results with p-values <0.05

29. Table 4: The authors need to describe the decision to classify “it’s up to god, or fatalism” as an effect modifier and the rationale for excluding it from the analysis.

Response: There was an error in our definition of the asterisk as an effect modifier which has since been revised. The * implied that the data on reason (It’s up to God or fatalism) was not collected in the 2003 survey.

30. Table 5 would be more accurately titled something like “Difference in method continuation rates by reason, among women in need of contraception in Kenya between 2003 and 2014.”

Response: The title of this table has been revised to reflect the proposed change.

31. Discussion:

Line 240: “A declined in contraceptive discontinuation between 2008 and 2014 was observed” seems to belie the preceding statement, lines 237-238, stating that no significant change was observed between 2003 and 2014—there are no statistical tests shown that support those statements.

Response: This has been revised based on the additional trend analysis that was conducted that indicates a significant change was observed (lines 264-265)

Submitted filename: Response to reviewers.docx

Click here for additional data file.

12 Oct 2020

PONE-D-20-19927R1

Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003–2014

PLOS ONE

Dear Dr. Ontiri,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Nov 26 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Catherine S. Todd

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

My thanks to the authors for responding to most of the reveiwers' comments. Please kindly address the minor edits requested by Reviewer 1 and this manuscript will then be acceptable for publication.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have adequately addressed all of both Revieweers' concerns in the text and additions to the tables.

Reviewer #2: I am largely satisfied with the authors’ changes and appreciate their efforts to respond to the reviewers’ comments so conscientiously. I note a few suggested clarifications for the authors.

On page 5, line 105, the authors refer to the method mix, but it appears that everything in lines 101-108 is prevalence of each method (the percentage of women using each method) rather than method mix (the percentage of use contributed by a particular method).

Optional: Also in this section, the authors not trends that are both substantial and likely statistically significant (e.g. increase of injectable use from 14.3% to 21.6% in 2003-2008/9) and those that are not (e.g. decrease of pill use from 7.5% to 7.2%). If the authors would like to streamline the paper, this is one paragraph where some modest edits could help achieve that.

On page 7, lines 149-150, the authors are technically correct that discontinuation while in need comprises women who discontinue while still at risk of becoming pregnant, do not want to become pregnant… However, isn’t this definition operationalized from the reasons for discontinuation (given in the second column of the calendar)? It would be helpful to clarify this and list which reasons comprise discontinuation while in need (e.g. husband disapproval, side effects, access, cost, wanted more effective method…) and those that do not comprise while in need (wanted to become pregnant, husband away/infrequent sex,…).

Thank you for clarifying that percentages in Table 4 do not sum to 100% because the table excludes discontinuation due to no further need. Could the table titled be revised to read, “Reasons for discontinuation while still in need” rather than “…among women in need,” as this change would be more accurate? And could the authors add a footnote to the tables similar to their response to the reviewer, e.g. “Percentages do not sum to 100% since women who discontinued while not in need of contraceptives, including those who reported they wanted to become pregnant/experienced menopause (2003, 42.1% and in 2014, 46.5%) were excluded from the analysis.”

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Kerry MacQuarrie

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

14 Oct 2020

Comments

1. On page 5, line 105, the authors refer to the method mix, but it appears that everything in lines 101-108 is prevalence of each method (the percentage of women using each method) rather than method mix (the percentage of use contributed by a particular method).

Response: This is noted. We have since updated the figures from line 98-106 and we are now reporting contraceptive method mix.

2. Optional: Also in this section, the authors not trends that are both substantial and likely statistically significant (e.g. increase of injectable use from 14.3% to 21.6% in 2003-2008/9) and those that are not (e.g. decrease of pill use from 7.5% to 7.2%). If the authors would like to streamline the paper, this is one paragraph where some modest edits could help achieve that.

Response: We take note of this comment. Ascertaining whether the changes observed in the trends of the method mix are statistically significant, is beyond the scope of our paper. We have however revised the write up to report on the changes observed without using the terms increase or decrease.

3. On page 7, lines 149-150, the authors are technically correct that discontinuation while in need comprises women who discontinue while still at risk of becoming pregnant, do not want to become pregnant… However, isn’t this definition operationalized from the reasons for discontinuation (given in the second column of the calendar)? It would be helpful to clarify this and list which reasons comprise discontinuation while in need (e.g. husband disapproval, side effects, access, cost, wanted more effective method…) and those that do not comprise while in need (wanted to become pregnant, husband away/infrequent sex,…).

Response: This is noted. We have since updated the write up to reflect the suggestion in lines 148-153

4. Thank you for clarifying that percentages in Table 4 do not sum to 100% because the table excludes discontinuation due to no further need. Could the table titled be revised to read, “Reasons for discontinuation while still in need” rather than “…among women in need,” as this change would be more accurate? And could the authors add a footnote to the tables similar to their response to the reviewer, e.g. “Percentages do not sum to 100% since women who discontinued while not in need of contraceptives, including those who reported they wanted to become pregnant/experienced menopause (2003, 42.1% and in 2014, 46.5%) were excluded from the analysis.”

Response: We have revised the title of the table as advised. We have also added a footnote.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

19 Oct 2020

Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003–2014

PONE-D-20-19927R2

Dear Ms. Ontiri,

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,

Catherine S. Todd

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

I thank the authors for their attentive revisions and am pleased to accept this manuscript for publication.

Reviewers' comments:

27 Oct 2020

PONE-D-20-19927R2

Patterns and determinants of modern contraceptive discontinuation among women of reproductive age: Analysis of Kenya Demographic Health Surveys, 2003–2014

Dear Dr. Ontiri:

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

Dr. Catherine S. Todd

Academic Editor

PLOS ONE