Trend and determinants of quality of family planning counseling in Ethiopia: Evidence from repeated PMA cross-sectional surveys, (2014-2019).

INTRODUCTION: The modern contraceptive prevalence rate (mCPR) among married women has increased by nearly five-fold in Ethiopia from 8.1% in 2000 to 37% in 2019. Despite this increase, receipt of high quality contraceptive counselling, as measured by the percentage of contraceptive users who were told about other methods, counseled on side effects and counseled what to do in the event that they encountered side effects, has declined in recent years. The quality of family planning counseling service measured by using these three components, known as the Method Information Index(MII), is an index designed to measure quality and a key indicator of the FP2020 initiative. The effects of potential client and service provider-level factors on receipt of high quality counseling and its progress over time have not been well studied in Ethiopia.
METHODS: We pooled data from seven Performance Monitoring for Action (PMA), formerly PMA2020, survey-rounds to examine the trend and effect of potential factors on receiving high quality of family planning counseling service in Ethiopia. Data from a total of 15,597 women aged 15 to 49 from seven survey-rounds were used in the analysis. To account for the study design and unequal probabilities of selection from target-populations for sampled women, design-based analysis was used to compute proportions. Multilevel ordinal regression model with enumeration area as a second level were employed to examine potential factors associated with quality of family planning counseling service.
RESULTS: We found that the percentage of women who received high quality family planning counseling service declined from 39% (95%CI: 33%, 44%) in 2015 to 12% in 2019 (95% CI: 10%, 14%) nationally. Amhara region had the lowest percentage of women receiving high quality counseling at both the earliest(2014) and latest(2019) survey rounds(17% and 6%, respectively). Results show that lack of media exposure about family planning, having no formal education, using short-acting methods, and getting the service from pharmacy were the main factors associated with receiving low quality family planning counseling service.
CONCLUSIONS: Given the importance of continuous provision of information on the range of family planning methods, it is imperative to use media and particularly regional media which can effectively address the rural populations in local languages as an important vehicle of information on family planning. Interventions aimed at improving quality of family planning counseling need to be mindful of regional disparities in the severity of the problem to ensure equity in service access. To improve the coverage of high quality family planning counseling service, there is an urgent need to re-visit the format of family planning counselling services.

1 Background

The 2030 Sustainable Development Goals (SDGs) calls on countries “by 2030, to ensure universal access to sexual and reproductive health-care services, including for family planning, information and education, and the integration of reproductive health into national strategies and programmes” [1]. In 2019, among 1.9 billion women of reproductive age (15–49 years) worldwide, 1.1 billion had a need for family planning [2]. Of these 842 million were users of modern contraceptive methods and 190 million women who wanted to avoid pregnancy were not using any contraceptive method [1]. In sub-Saharan Africa, annual rate of changes (percentage points) in modern contraceptive use among women of reproductive age varied substantially from 0.77 to 3.17 [3].

Counseling plays a key role in enhancing family planning services, providing contraception related information and supporting family planning and fertility goals for women of childbearing age [4]. The Method Information Index (MII) is the current standard proposed by the Track 20 initiative to monitor quality of family planning counseling [5, 6]. The simple index uses three questions—whether the respondent received counseling on multiple contraceptive methods, whether she was told about side-effects, and if so, whether she was told what to do about side-effects—that are available in the majority of population-based surveys, including the Demographic and Health Survey (the DHS) and Performance Monitoring for Action (PMA) [7]. Adaptations have been proposed to the MII, including a fourth question assessing whether women were told they could switch to another method. This question was added in PMA 2019 Survey, however, and for trend purposes is not included in the current analysis.

Despite substantial global efforts to improve conceptualization, measurement, and delivery of high-quality family planning services, studies consistently find that receipt of high-quality services is low. In a study exploring the levels and trends of the MII across 25 countries, urban residence, education, household wealth, and method type were generally found to be associated with receiving higher quality care [8]. In all countries combined, the median MII values show that the contraceptive information received by women varied by method type and was highest for implant users and lowest among women relying on sterilization.

In Ethiopia specifically, despite increases in contraceptive coverage over time the percentage of contraceptive users who received all components of Method Information Index(MII) has not changed substantially [9, 10]. Studies on family planning counseling services in Ethiopia consistently show that high quality of family planning counseling service coverage is below 40% [11-14].

Understanding which individual, social, and health system factors are associated with the receipt of high-quality family planning counseling is critical in order to identify inequities and improve overall service delivery. Multiple studies have assessed facility and provider characteristics that are associated with the receipt of quality family planning counseling services [15-18], but relatively few studies have assessed the client-level and social factors that may be related to receipt of high quality counseling service.

A previous study in Ethiopia found that in addition to these factors, women who received family planning services from a public provider had higher odds of receiving all components of the MII. Little additional evidence exists, however, on the client-level factors that may be associated with receiving high quality services, resulting in a call for additional research into identifying factors associated with quality [19].

Studies that have explored client-level and social factors associated with higher quality services have generally been limited by use of one to two cross-sectional surveys which limit the ability to detect trends in receipt of high quality family planning counseling service over time. Increasingly, however, it is common to see repeated cross-sectional surveys carried out in multiple rounds over short periods of time, particularly surveys conducted by the Performance Monitoring for Action (PMA) project. Pooling such data improves the power of test due to an increased sample size and can be used to better investigate trends over-time. A study done by [12] used pooled data from five survey rounds in Ethiopia to compute the percentage of women who received high counseling service across survey years by different factors. However, the analysis on factors of quality of family planning counseling service was limited only to one survey round (i.e., 2018 survey). To our knowledge, there is no study that considered repeated cross-sectional surveys to identify factors related to quality of counseling service in Ethiopia.

In the present study, we use repeated cross-sectional data from seven-rounds of a national survey to: i) establish the trends in quality of family planning counseling among all women aged 15–49 years, ii) identify client-level and social factors associated with quality of counseling, and iii) identify regional disparities in quality of counseling. Findings from this study can help to inform planning interventions to improve the quality of family planning counseling services in Ethiopia.

2 Materials and methods

2.1 Study design and data source

The data for this study were obtained from seven rounds of the Performance Monitoring for Action (PMA) surveys, which were undertaken yearly since 2014 (formerly as Performance Monitoring and Accountability 2020 (PMA2020)). The PMA surveys are repeated cross-sectional surveys based on a multistage stratified cluster sampling design with urban-rural stratification for Amhara, Oromia, Tigray, and Southern Nations, Nationalities, and Peoples’ (SNNP) regions, and for the remaining regions, regions served as the strata, without additional urban/rural stratification. The primary sampling units or Enumeration Areas (EAs) are selected using probability proportional-to-size method for which the sample selection probability depends on the size of population. For the first round (referred to as PMA2014/Ethiopia) and latest round considered in this study (referred to as PMA2019/Ethiopia) of data collection the survey conducted in 200 and 265 enumeration areas, respectively, which were selected by the Central Statistics Agency (CSA) to be representative at the national and regional level. Prior to data collection, all households in each enumeration area were listed and mapped by the resident enumerators to create a frame for the second stage of the sampling process. This mapping and listing process was done in the first week of data collection in each enumeration area. Once listing was completed, 35 households were randomly selected by field supervisors using a phone-based random number-generating application. All occupants in selected households were enumerated and from this list, all eligible women between the ages of 15 to 49 were administered the female-household questionnaire after informed consent was obtained. In PMA surveys considered in this analysis, the response rate is high across different survey years which ranges from 95.6% to 99.2%. In all survey-rounds, data were collected using smartphones.

The details of the PMA2020 survey design and sampling techniques are described elsewhere [20], while updates made to the sampling design in 2019 are further described by Zimmerman and colleagues [21]. PMA uses standard data collection procedures which are consistent across-rounds, and consistent in content, allowing pooling the data and making comparisons over time. To assess trend over time and identify factors linked with the outcome variable, described below the pooled data from seven-rounds, spanning 2014 to 2019 were considered.

2.2 Sample

In this study, data from nationally representative cross-sectional female and household surveys collected on yearly bases from 2014 to 2019 among women aged 15–49 were utilized. A total of 15,597 women who were asked about the three questions, described below, were considered as our analytical sample.

2.3 Variables

The outcome variable in this study, quality of counseling service, was derived based on the responses to the three equally weighted questions used in the MII. Women were asked to report whether, at the visit where they first received their current contraceptive method, they were: i) told about other methods, ii) counseled on side effects, and iii) counseled on what to do if she experienced side effects. Originally, the data related to the variable of interest were collected using possible Yes or No questions to assess quality of counseling. Using these questions, the MII was assessed two separate ways; first, as a binary variable, indicating whether a woman received all components of the MII (“high-quality”) or not, and second, as an ordinal variable. We defined the ordinal variable with four categories indicating “high” quality (all three components received), “moderate” quality (any two of three), “low” quality (any one of three), and “no” counseling (no components received).

In order to assess the effect of different factors on receiving high quality family planning counseling service, the outcome variable for the i women from j enumeration area(EA), Y, is dichotomized as follows:

This categorization was used to compute design-based prevalence of high quality family planning counseling service by different socio-demographic factors, described further below.

The analysis using the dichotomized outcome cannot provide information about different counseling levels simultaneously, thus further disaggregation was done based on the following categorization:

After reviewing related literature, possible factors linked with the quality of family planning counseling were extracted from the survey data. Table 1 presents the list of factors we hypothesize are associated with the receipt of quality of family planning counseling service considered in this study. We define long-acting methods as female sterilization, implant and IUD and short-acting methods as the injectable, pill, emergency contraception, male condom, the Standard Days Method, and lactational amehorrhea method (LAM).

Table 1

Description of exploratory variables considered in the study.

Variable	Variable description and/or categorization
Age	Classified as: 15–19 20–24, 25–34, 35–49
Education	Never attended, Primary, Secondary/higher
Parity	Classified as: 0 children, 1–2 children, 3+ children
Method type	Categorized as: long-acting and short-acting
Wealth	Household wealth quintile (lowest to highest)
Media exposure†	Exposure to family planning information disseminated through media
Method source	Source of method (hospital, health post, health center, pharmacy, other)
Residence	Place of residence: urban, rural
Region	Tigray, Amhara, Oromia, SNNP, Addis Ababa, Other‡
Survey year¶	Survey year considered as categorical variable

† Composite variable constructed using YES or NO question: have you heard/seen or received message about family planning from radio, TV, newspaper or magazine, social media, text message on mobile.

‡ Other regions represents the six small regions, namely Afar, Somali, B-Gumuz, Gambella, Harari and Dire Dawa.

¶ Round 1 and 2 surveys were done in 2014.

2.4 Statistical analysis

Nationally representative survey data-sets are often complex in nature for two reasons: i) the use of stratified multistage cluster sampling, and ii) unequal probabilities of selection from target-populations for sampled elements, often as a result of oversampling of key subgroups. Thus to account for the study design and unequal probabilities of selection from target-populations for sampled women, sampling weights were used during the estimation of proportions and their respective confidence interval [22].

The types of models used for data analysis depend on the nature and measurement scale of the outcome variable under consideration. The level of measurement of the variable is both binary and ordinal. When the response categories are ordered, the use of this ordering yields more parsimoniously parameterized models [23]. In such models, the ordinal nature of the response is taken into account by considering the cumulative probabilities. The data from different women are assumed to be independent, but due to the clustered nature of the data, observations from the same cluster/EA may not be independent.

Women living in the same EA may share similar health facilities, and the quality of counseling services available to them may be more alike than women living in another EA. Hence, our modeling approach should takes into account the correlated nature of the data by introducing a random effects term to account for a variety of situations, including EA heterogeneity, and unobserved covariates within that EA. As a result, in this study we employ a random effects model for ordinal outcomes.

Let μ be the probability of the i subject from the j EA being in the response category k, μ = P(y = k). Further, we let cumulative probability of the response in category k or above represented by Π = P(y ≥ k). The random intercept proportional odds model (POM) is given by: where k is the level of the ordered category. The parameter β0 is the intercept for category k, usually is considered as nuisance parameter of little interest ([23]), X is a vector of exploratory variables described in Table 1 with associated regression coefficients β. The random intercept b is assumed to vary randomly among EA’s according to a distribution. The variance of the random effect () represents how much variability there is between clusters with respect to the response variable.

Model 2 assumes an identical effect of the predictors for each cumulative probability. The model assumes that, conditional on the random effects, the coefficients that describe the relationship between, for example, the lowest versus all higher categories of the response variable are the same as those that describe the relationship between the next lowest category and all higher categories, etc. This is called the proportional odds assumption or the parallel regression assumption [23]. Specifically, the model implies that odds ratios for describing effects of explanatory variables on the response variable are the same for each of the possible ways of collapsing the response to a binary variable. The proportional odds (PO) assumption for each variable were tested using the Brant test ([24]), and held across all variables.

Further, to quantify the relative importance of clusters as a source of variation about receiving high counseling service, EA-level random effects variance was expressed in terms of Variance Partition Coefficients (VPC).

The proportion of total variation of receiving high family planning counseling attributable to EA random effect () is computed as follows [25, 26]: where is EA (level-2) random effect variance, and π2/3 = 3.29 is individual level variance is equal to the variance of a logistic distribution [27]. In Model 2, fixed effects parameter has a conditional interpretation. It refers to the consequence of changing the value of an explanatory variable, for which the fixed effect is the coefficient, for a given value of the random effect and the other fixed effects.

In all exploratory data analyses, sample weights calculated based on the multistage sampling design were considered. Data analyses were done using Stata 16.1 software [28], and graphs presented in this manuscript were generated using the R software [29] by ggplot2 package [30]. In this study, for statistical significance, the considered alpha(type-I error) is 0.05.

2.5 Ethics approval statement

The study protocol was approved by Institutional Review Boards of College of Health Science at Addis Ababa University, Ethiopia and at the Bloomberg School of Public Health at Johns Hopkins University in Baltimore, USA.

3 Results

3.1 Descriptive results

Table 2 presents client, service provider, and regional level characteristics of the sampled study population by year. Across all years, the majority of women lived in rural areas. Approximately 50% of respondents were age 25–34 years, and less than 20% of respondents were nulliparous. On average, around 50% of women got received their contraceptive methods from health centers. The use of long-acting contraception methods (female sterilization, implant and IUD) increased from 21.81% in 2014 to 36.94% in 2019. The percentage of respondents who reported hearing a message about family planning from the media decreased across survey years.

Table 2

Sample characteristics of PMA survey respondents across survey years.

Factor		Survey Round
		2014 †	2015	2016	2017	2018	2019
Total (n)		3889	2450	2503	2366	2353	2036
Residence	Urban	40.11	40.86	42.67	42.39	43.43	45.38
	rural	59.89	59.14	57.33	57.61	56.57	54.62
Region	Tigray	17.82	14.94	13.62	14.45	14.28	11.59
	Amhara	24.79	20.98	22.97	21.89	22.95	22.20
	Oromia	12.81	19.71	19.50	19.32	20.82	22.20
	SNNP	26.10	26.12	26.01	24.43	24.14	20.43
	Addis Ababa	12.45	12.12	11.91	11.88	11.64	9.72
	Other	6.04	6.12	5.99	8.03	6.16	13.85
Education	Never attended	32.57	30.73	30.78	29.24	28.07	32.12
	Primary	37.85	37.19	37.73	36.48	36.80	39.15
	Secondary/higher	29.58	32.08	31.49	34.28	35.14	28.73
Age(year)	15–19	7.02	7.35	7.03	8.62	6.84	7.66
	20–24	22.68	24.86	22.41	22.36	20.95	20.97
	25–34	47.06	45.10	46.66	45.73	47.22	46.66
	35–49	23.25	22.69	23.89	23.29	24.99	24.71
Parity	0 Children	15.66	18.82	15.26	17.97	17.00	10.76
	1–2 children	42.76	41.67	44.19	43.38	42.07	46.39
	3+ children	41.58	39.51	40.55	38.65	40.93	42.85
Method source	Hospital	17.26	17.51	16.42	20.63	21.76	9.04
	Health post	19.64	19.28	17.80	20.45	18.53	22.79
	Health center	50.42	48.86	52.4	44.96	43.56	60.17
	Pharmacy	6.62	8.59	7.96	9.19	11.39	6.43
	Others	6.06	5.77	5.42	4.77	4.76	1.57
Method type	Short-acting	78.19	75.21	73.79	72.03	69.62	63.06
	Long-acting	21.81	24.79	26.21	27.97	30.38	36.94
Media exposure	No	37.80	36.32	38.50	44.14	44.26	53.93
	Yes	62.20	63.68	61.50	55.86	55.74	46.07
Wealth quintile	Lowest	7.02	8.9	8.99	9.55	9.82	11.89
	Lower	9.39	9.22	10.31	9.86	10.84	15.28
	Middle	11.03	10.86	12.78	10.23	10.62	17.83
	High	17.56	17.76	16.74	20.84	20.7	20.87
	Highest	55	53.27	51.18	49.54	48.02	34.14

† In 2014 two surveys were conducted, and the data were pooled together.

The overall weighted prevalence of receiving high quality family planning counseling service was 28.61% (with 95% CI: 27.68%—29.55%). Table A1 in S1 Appendix presents the percentage distribution of women who received each level of family planning counseling quality by different factors and over the survey-rounds. The percentage of women who received high quality family planning counseling service was 27.6% in 2014, increased to 38.6% in 2015, and thereafter declined to a low of 11.8% in 2019 (Table A1 in S1 Appendix, Fig 1). Receiving a method from a health center or health post, using a long-acting method, and having media exposure to family planning were associated with a higher prevalence of receiving high quality family planning counseling service across all survey rounds.

Fig 1

Percentage of women receiving different levels of family planning counseling service over survey years.

Table 3 presents a summary of the percentage change in receiving high quality family planning counseling service. Compared with 2018, there was a significant decline (by 59.4%) in the coverage of providing high counseling service in 2019. Overall, the percentage of the women who received high quality family planning counseling service declined between 2014 and 2019 by 57.3 percentage points.

Table 3

Percentage change in receipt of high-quality family planning services across survey rounds.

Base year	Survey year
	2015	2016	2017	2018	2019
2014	39.9(19.57, 60.25)	16.6(-3.03, 36.2)	14.3(-8.52, 37.2)	5.1(-16.1, 26.3)	-57.3(-67.11, -47.48)
2015		-16.7(-28.10, -5.27)	-18.3(-34.7, -1.9)	-24(-40.1, 9.7)	-69.47(-76.5, -62.4)
2016			-2.0(22.02, 18.17)	-9.84(-28.44, 8.7)	-63.4(-71.9, -54.8)
2017				-8.1(-21.84, 4.69)	-62.65(-71.05, -54.25)
2018					-59.4(-68.56, -50.18)

The percentage of high quality family planning counseling changed over time with different rate across regions of the country (Fig 2, Table A1 in S1 Appendix). The coverage in the receipt of high quality of counseling in Tigray, Addis Ababa and SSNP region was above the national level while, the coverage was found to be lower in Amhara region (Fig 2).

Fig 2

Percentage of women receiving high quality family planning counseling service over survey years by region.

Fig 3 presents overall trends in the percentage of women who received high quality family planning counseling service across survey year by different factors. The results reveal that, irrespective of the factors examineds, the quality of counseling increased from the year 2014 to 2015 and start declining thereafter.

Fig 3

Trends in receipt of high quality family planning counseling by different factors.

Percentage of women who received high family planning counseling service according to: a) type of method used, b) exposure to family Planning via media, and c) source of contraceptive method.

3.2 Modeling results

Table 4 presents the multilevel modeling results by considering one factor at a time (Model A) and after controlling for observed explanatory variables (Model B). Conditional on EA level random effects, the odds of getting counseling service varied across categories for parity, method type, education, media exposure, method source, region of residence, and survey year.

Table 4

Parameter estimates from the multilevel ordinal logistic regression model of receiving high quality family planning counseling service in Ethiopia.

Factor	Model A		Model B
	COR	COR 95% CI	AOR	AOR 95% CI
Age (ref:15–19)
20–24	1.31	(1.11, 1.54)	1.01	(0.80, 1.24)
25–34	1.64	(1.39, 1.93)	1.19	(0.93, 1.53)
35–49	1.65	(1.37, 1.98)	1.18	(0.90, 1.56)
Wealth Quantile (ref: Lowest)
Lower	0.88	(0.74, 1.05)	0.89	(0.74, 1.07)
Middle	1.02	(0.85, 1.22)	1.03	(0.84, 1.26)
Higher	1.08	(0.90, 1.30)	1.05	(0.85, 1.29)
Highest	1.23	(1.01, 1.49)	1.03	(0.80, 1.33)
Parity (ref: 0 children)
1- 2 children	1.77	(1.56, 2.01)	1.36	(1.14, 1.63)
3+ children	1.89	(1.65, 2.17)	1.47	(1.16, 2.14)
Method type (ref: short acting)
Long-acting	2.29	(2.04, 2.57)	1.90	(1.69, 2.14)
Education (ref: Never attended)
Primary	0.99	(0.89, 1.09)	1.07	(0.95, 1.21)
Secondary/higher	1.16	(1.02, 1.32)	1.38	(1.17, 1.62)
Media exposure (Ref: No)
Yes	1.75	(1.59, 1.93)	1.68	(1.51, 1.88)
Method source (ref: Hospital)
Health post	1.82	(1.53, 2.16)	2.18	(1.80, 2.63)
Health center	2.02	(1.74, 2.36)	2.15	(1.82, 2.54)
Pharmacy	0.23	(0.18, 0.30)	0.29	(0.22, 0.38)
Other	1.01	(0.76, 1.35)	1.09	(0.83, 1.42)
Residence (ref: Urban)
Rural	0.84	(0.72, 0.98)	0.99	(0.80, 1.22)
Region (ref: Addis Ababa)
Tigray	0.86	(0.65, 1.13)	0.78	(0.61, 0.99)
Amhara	0.44	(0.34, 0.57)	0.47	(0.37, 0.59)
Oromiya	0.69	(0.53, 0.89)	0.67	(0.53, 0.84)
SNNP	0.92	(0.71, 1.19)	0.88	(0.69, 1.13)
Other	0.69	(0.51, 0.93)	0.75	(0.53, 1.07)
Survey year (ref:2014)
2015	1.41	(1.10, 1.79)	1.53	(1.15, 2.03)
2016	1.26	(1.02, 1.56)	1.25	(1.01, 1.55)
2017	1.07	(0.84, 1.35)	1.11	(0.86, 1.42)
2018	0.98	(0.79, 1.23)	1.02	(0.81, 1.27)
2019	0.54	(0.44, 0.67)	0.43	(0.35, 0.54)
Random effect
Var(EA)	0.71		0.58	(0.47, 0.73)
VPC	21.58%		17.63%

Note: Model A is fitted by considering only one predictor variable at time, and the variance of the random intercept varies from 0.60 to 0.81 with an average variance of 0.71. Model B is fitted by adjusting for client-level, service provider-level, and place of residence variables simultaneously. VPC represents the percentage of total variation of receiving high family planning counseling attributable by EA

Women who reported being exposed to family planning information through the media, those who received their family planning method from health post or health center, women who attained secondary or higher education, and women using long-acting methods were significantly more likely to receive high counseling service on family planning (Table 4).

Among women with the same random effect value, relative to women who were using short-acting methods, women who were using long-acting methods had 1.90 times the estimated odds of getting “high” counseling service instead of “moderate” or “low” or “no” counseling and the estimated odds of getting “high” or “moderate” counseling service instead of “low” or “no” counseling service and the estimated odds of getting “high” or “moderate” or “low” counseling service instead of “no” counseling service. Among women with the same random effect, compared to women who got their method from a hospital, women who got their method from a health post had 2.18 times the estimated odds of getting “high” counseling service instead of “moderate” or “low” or “no” counseling service and the estimated odds of getting “high” or “moderate” counseling service instead of “low” or “no” counseling service and the estimated odds of getting “high” or “moderate” or “low” counseling service instead of “no” counseling service (Table 4).

For Model B (Table 4), the variance partition coefficient (VPC) is , which indicates that 17.63% if the variance in counseling service can be attributed to differences between EA’s.

4 Discussion

This study examined factors associated with receipt of high quality family planning counseling services and its trend in Ethiopia from 2014 to 2019. We found that the percentage of women who received high-quality services was low throughout the study period, reaching a high of 38.6% in 2015 and declining sharply to 11.8% in 2019. Significant variation in receipt of high quality counseling was found at the client-level, service delivery point-level, and by region of residence. Nationally, our finding on the trend analysis in the percentage of women receiving high quality family planning counseling service declined overtime. This findings point out some of the gains in improving service coverage were not complimented with improved service quality. In addition, the fact that exposure to family planning message from the media and health professionals has decreased over the years indicate perhaps less attention was given to maintaining and improving the existing public awareness on family planning methods. While the level of high-quality counseling did not exceed 40% at any point, there was a sharp decline in 2019, falling almost 60% from the year before. Among client-level factors, parity, education, media exposure, and type of contraceptive method were found to be statistically significantly associated with high quality family planning counseling services.

Except in Addis Ababa, high-quality family planning counseling peaked in 2015 in all other regions and declined in the following years. The observed relatively better family planning counseling service in 2015 may be due to the fact the Ethiopian health sector transformation plan started in this year [31]. As compared with other regions, high quality counseling service was low in Amhara region and sharply declined since 2017 (Fig 2). This aligns with Hrusa’s finding using PMA data [12].

At the client-level, Jain found that education and wealth were generally associated with higher quality services [8]. We found no differences by wealth, though we did see that women who attended secondary or higher education were significantly more likely to report receiving higher quality services than women without education. Similarly, women who heard about family planning in the media were also more likely to indicate they received higher quality counseling services. These results provide support for Jain’s hypothesis that women with more education may have greater method knowledge and greater ability to recall the information exchanged during their family planning visit. Engagement with family planning media may also lead to greater knowledge of different contraceptive methods and their side effects, allowing women to ask questions of providers that may otherwise be overlooked. Continued training to ensure that comprehensive contraceptive information is delivered in simple, understandable language may alleviate some educational disparities, as would investment in ongoing family planning media campaigns. Messaging that is developed in conjunction with regional media, and distributed in local languages, may support greater general knowledge of family planning and support women in asking questions and receiving high quality services. Media exposure enables a more informative client-provider interaction because it’s enhances client knowledge and empowers them to be more inquisitive in the family planning visit.

In line with findings in other countries, education, parity have impact on receiving high family planning counseling service [8, 15, 32–34]. A study by [14] showed, 58.9% of the counseling sessions did not maintain the privacy of clients during the consultation and 74.2% of them were not told about the possible side effects related to the use of a method. A study done in the northern part of Ethiopia shows that only 22.4% of the clients chose contraceptive method by themselves and 74.1% of them with assistance of the provider [13]. Variation in access to and quality of family planning services by parity and method-type has been documented extensively across settings and has identified consistent barriers to high quality family planning services among unmarried adolescents, including provider bias and lack of training [14, 35, 36]. Policies and programs designed to expand and strengthen youth friendly services must emphasize that the delivery of comprehensive information is a critical component of high-quality youth friendly services, in addition to improving privacy, confidentiality, and reducing provider bias.

Finally, women who reported using a long-acting method were significantly more likely to report receiving high quality services than women using short-acting methods. This aligns with Jain’s findings that the MII was highest among implant users, followed by the IUD [8]. While it is positive that women who choose long-acting methods are made aware of potential side effects and how to address them, the majority of contraceptive users in Ethiopia use the injectable [9]. Even as the method mix changes in Ethiopia, with greater use of long-acting methods, it is important to ensure that women who continue to prefer short-acting methods are given comprehensive information.

At the service provider-level, method source was significantly associated with quality of family planning counseling service. Women who received care from pharmacies had significantly lower odds of receiveing high-quality family planning counseling than women who received care at either a health post or health center. In addition to differences in service modality and training of service providers, this likely reflects differences in characteristics of users and method choice; other studies have shown that women who use pharmacies tend to live in urban areas, be younger, and rely on short-term methods such as emergency contraceptives and condoms which are typically not preceded by sufficient counseling [37, 38].

We did not find any differences by urban or rural residence, which is contrary to the recent Hrusa study in Ethiopia that found rural women were more likely to receive high quality services [12]. This may be due to the fact that the paper did not account for clustering at the EA level, which may be particularly important among rural women who have fewer health facilities available to them than urban women. While Jain found that across 25 countries, urban women were more likely to receive high quality counseling, Ethiopia has invested heavily in its public health system, particularly in rural areas, to address disparities in access to and use of family planning services [31]. This, combined with the generally higher level of services available at public health facilities (i.e., health post and health centers), may explain why urban and rural differences are not pronounced.

The results presented in this study should be considered in light of some limitations. One limitation is the fact that the MII does not capture the full range of elements necessary to measure high-quality contraceptive counseling. Additionally, due to the nature of the household survey which asks women to recall information about the first time they received their current method, there may be a recall bias. Further, covariates considered in this study were limited to those available in the survey which lacks supply-side variables such as availability of methods at facilities, availability of trained staff, job aids, information leaflets for clients and other facility level variables. Since the analysis results in this study were derived from cross-sectional female and household surveys, and quality of counseling is heavily influenced by facility-level factors, future studies should consider using PMA service delivery point (SDP) surveys.

5 Conclusions and recommendation

Quality of family planning services is low in Ethiopia, with evidence of a sharp decline in recent years. Significant disparities by region, method source and by socio-demographic characteristics exist, particularly among uneducated women, media exposure about family planning and by choice of method. Given the importance of continuous provision of information on the range of family planning methods and their side effects, it is imperative to develop simple, culturally appropriate messages in local languages and collaborate with regional media as an important vehicle of information on family planning. Interventions aimed at improving quality of family planning counseling need to be mindful of the regional disparities and ensure equity in service quality. Further, the results of this study reveal that high-quality family planning counseling service tended to be more available at health posts and health centers compared to hospitals and is particularly poor at pharmacies. Thus, the Ministry of Health and partners should support training on contraceptive counseling for pharmacists and related auxiliary health staff to improve the information provided through the private sector.

(PDF)

Click here for additional data file.

31 Aug 2021

PONE-D-21-19351

Trend and Determinants of Quality of Family Planning Counseling in Ethiopia: Evidence from repeated PMA cross-sectional surveys, (2014-2019)

PLOS ONE

Dear Dr. Ejigu,

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 Oct 15 2021 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 . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at  https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Orvalho Augusto, MD, MPH

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

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

"This study was conducted with support received from the Bill & Melinda Gates Foundation through a grant received by the Bill & Melinda Gates Institute for

Population and Reproductive Health for the Performance Monitoring for Action Ethiopia (PMA-Ethiopia) projects."

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

"This study was conducted with support

received from the Bill & Melinda Gates

Foundation through a grant received by the Bill & Melinda Gates Institute for

Population and Reproductive Health for the Performance Monitoring for Action

Ethiopia (PMA-Ethiopia) projects."

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. Please provide additional details regarding participant consent. In the ethics statement in the Methods and online submission information, please ensure that you have specified (1) whether consent was informed and (2) what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed). If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information.

If you are reporting a retrospective study of medical records or archived samples, please ensure that you have discussed whether all data were fully anonymized before you accessed them and/or whether the IRB or ethics committee waived the requirement for informed consent. If patients provided informed written consent to have data from their medical records used in research, please include this information.

Additional Editor Comments (if provided):

Family planning is such a very important service. Here the authors report a longitudinal analysis of quality counselling using PMA surveys. I applaud the authors for first, use longitudinally a series of 7 cross-sectional datasets, secondly for defining just one ordinal outcome rather than making 3 logistic regression (as many would do). This is a very efficient use of data. However, it is still debatable to use just these 3 variables as indicators of the quality of counselling. I see no discussion of this.

Issues

I. Abstract:

Please split this into introduction, methods, results and conclusions

II. Introduction

This is a very good background. Just a clarification on line 12 what are these numbers (0.77 to 3.17)? Are these yearly changes?

III. Methods

1. In the “study design and data source” we are pointed to reference 26 to learn the sampling details of PMA2020. That is fine but brief details of such sampling should be added here. It would be great if details of what is the size of an Enumeration Area or of the PSU relative to an administrative area in Ethiopia. How is comparable?

2. What is this method of information index? (lines 77/78)

3. Please add spaces at “if a woman got allthethree” between lines 84 and 85

4. Equation 2 - nowhere is explained the meaning of sigma2 squared (the variance of the random-effects).

5. Table 1 - please either specify what “svy-year” means or write “survey year”.

6. Line 122 - the Brant is a name. So please correct it to be “Brant test”.

7. Line 136 - Stata is not an acronym. So correct STATA to Stata please. See the official documentation for reference. For example, your reference 34 clarifies that.

8. Line 137, I suggest adding the reference for ggplot2. Your plots are based on ggplot2. Correct?

9. Was any goodness of fitness performed in this analysis?

IV. Results

1. What is the amount of missing data here?

2. Table 2 please on the caption/footnote that these are proportions in percentages.

- Are these weighted proportions?

- all percentages seem to respect the 2 decimal places. So why not “2014 FP message” “2015 FP Facility”?

3. Table 3 there are empty cells. Such as 2019 other regions. Can you check this, please?

4. Line 166 no figure or table indicated.

5. How the confidence intervals for the relative changes are computed?

6. Table 5 - What are COR and AOR? Put in the footnote, please. Here should be the place to add the variance component proportion. In fact, I do not see this in the results.

V. Discussion:

Why no limitations discussion here?

[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

**********

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

Reviewer #1: No

**********

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: No

**********

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: No

**********

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: The author chose a very important topic and I admire her/his statistical knowledge and skills. I would very much like to see a paper emerge on this subject. However, I feel that the authors have missed the mark in several respects that I outline below. I recommend that the authors take my feedback into consideration and determine how best to proceed themselves. The description of the data was so inadequate that I would not have been able to make my way through the paper had I not been somewhat familiar with the data source myself; however, my feedback might have missed the mark also since I had to make assumptions about the data source. Still, I would not reject a paper on this topic with these data, taking into account what else I know about PMA 2020 data but I think the analysis presented here with cannot go forward without first presenting the data source and collection methods more clearly, more thoroughly conceptualizing the notion of FP counseling quality in the wider context of FP service quality and mapping the factors that influence it vis-à-vis the array of indicators in the data sources available, and then thinking through more soundly what factors are most feasible and relevant for their analysis given the data at hand given the possibility of biases.

**********

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: Yes: Colin Baynes

[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: pLOS review.docx

Click here for additional data file.

2 Dec 2021

Response to Editor and Reviewers

Editor Comments

Family planning is such a very important service. Here the authors report a longitudinal analysis of quality counselling using PMA surveys. I applaud the authors for first, use longitudinally a series of 7 cross-sectional datasets, secondly for defining just one ordinal outcome rather than making 3 logistic regressions (as many would do). This is a very efficient use of data. However, it is still debatable to use just these 3 variables as indicators of the quality of counselling. I see no discussion of this.

Issues

I. Abstract:

Please split this into introduction, methods, results and conclusions

The abstract section updated accordingly.

II. Introduction

This is a very good background. Just a clarification on line 12 what are these numbers (0.77 to 3.17)? Are these yearly changes?

- It is an annual rate of changes (percentage points) in modern contraceptive prevalence rates among all women of reproductive age (15–49 years).

III. Methods

1. In the “study design and data source” we are pointed to reference 26 to learn the sampling details of PMA2020. That is fine but brief details of such sampling should be added here. It would be great if details of what is the size of an Enumeration Area or of the PSU relative to an administrative area in Ethiopia. How is comparable?

- In the revised version of the manuscript, additional clarification in this issue was given.

2. What is this method of information index? (lines 77/78)

- It is a composite score for family planning service quality.

3. Please add spaces at “if a woman got allthethree” between lines 84 and 85

- Thanks for pointing this, updated accordingly.

4. Equation 2 - nowhere is explained the meaning of sigma2 squared (the variance of the random-effects).

- The variance of the random effect (\\sigma_{2}^{2}) represents how much variability there is between clusters with respect to the response variable.

5. Table 1 - please either specify what “svy-year” means or write “survey year”.

Updated.

6. Line 122 - the Brant is a name. So please correct it to be “Brant test”.

- Corrected.

7. Line 136 - Stata is not an acronym. So correct STATA to Stata please. See the official documentation for reference. For example, your reference 34 clarifies that.

- Corrected.

8. Line 137, I suggest adding the reference for ggplot2. Your plots are based on ggplot2. Correct?

Yes, and reference on ggplot2 included.

9. Was any goodness of fitness performed in this analysis?

- Yes. Likelihood ratio test was employed to assess the importance of incorporating the random effect term in the model. Further, AIC was used to compare models with different covariates.

IV. Results

1. What is the amount of missing data here?

- Our analytical samples are all women who completed the survey questionnaires. counseling service.

2. Table 2 please on the caption/footnote that these are proportions in percentages.

- Are these weighted proportions? Yes

- all percentages seem to respect the 2 decimal places. So why not “2014 FP message” “2015 FP Facility”? Corrected.

3. Table 3 there are empty cells. Such as 2019 other regions. Can you check this, please?

- Checked and addressed.

4. Line 166 no figure or table indicated.

- Addressed.

5. How the confidence intervals for the relative changes are computed?

- Using the linearization approach.

6. Table 5 - What are COR and AOR? Put in the footnote, please. Here should be the place to add the variance component proportion. In fact, I do not see this in the results.

- COR=Crude Odds Ratio, AOR= Adjusted Odds Ratio

V. Discussion:

Why no limitations discussion here?

- In the revised version of the manuscript, a limitation section included.

Reviewer #1:

The author chose a very important topic and I admire her/his statistical knowledge and skills. I would very much like to see a paper emerge on this subject. However, I feel that the authors have missed the mark in several respects that I outline below. I recommend that the authors take my feedback into consideration and determine how best to proceed themselves. The description of the data was so inadequate that I would not have been able to make my way through the paper had I not been somewhat familiar with the data source myself; however, my feedback might have missed the mark also since I had to make assumptions about the data source. Still, I would not reject a paper on this topic with these data, taking into account what else I know about PMA 2020 data but I think the analysis presented here with cannot go forward without first presenting the data source and collection methods more clearly, more thoroughly conceptualizing the notion of FP counseling quality in the wider context of FP service quality and mapping the factors that influence it vis-à-vis the array of indicators in the data sources available, and then thinking through more soundly what factors are most feasible and relevant for their analysis given the data at hand given the possibility of biases.

Abstract

• I would not refer to 2014 as “baseline” and 2019 as “endline” since that implies that there was an intervention in between the years.

• “A multi-level model was used to examine potential factors”. I would state your research question(s) related to exploring trends and associations more clearly and for each the corresponding statistical strategy you use to answer the question.

• What is the operational definition of “good family planning counseling” and “poor family planning counseling”? How did you construct it as a variable in the statistical analyses?

• For readers that do not know indicate briefly the type of data that you are using (i.e., household surveys, facility-based exit interviews).

- In the revised version of the manuscript, the abstract entirely updated based on the Editor and reviewer comments.

Background

“In sub-Saharan Africa countries the trend in using 10 modern contraceptive method among women of reproductive age varied substantially from 0.77 to 3.17” – I do not understand this sentence. What is meant by 0.77 to 3.17?

- It is an annual rate of changes (percentage points) in modern contraceptive prevalence rates among all women of reproductive age (15–49 years).

“Prevalence of family planning coverage” – What is this? Prevalence to me connotes the level of method use among individuals in a population and coverage implies that amount of population that has access to a service (two different things). Similarly in line 15 you say “coverage” of contraception is associated with individual-level covariates. Do you mean “use of contraception”?

- Yes, to mean “use of contraception”, and updated accordingly.

The author frames the background with the global situation of contraception prevalence and unmet need in broad terms and leaps to a discussion of very focused set of factors that influence method use, counseling. A more useful framing might start by quickly acknowledging gaps in availability and use of FP services among women in sub-Saharan Africa, and proceed to frame the paper around the need to better understand the quality of care – pointing out the multiple aspects of family planning service quality and a overview of the approaches that have been used to measure and evaluate FP service quality, evidence on the determinants and consequences of FP service quality, then introducing the subject of this paper – the determinants of FP counseling quality - in terms of the particular knowledge gap that it is filling.

- We greatly appreciate the feedbacks, and the background section updated.

Be sure to clearly define family planning counseling and how the quality of FP counseling has been defined and studied in the literature. There are multiple dimensions to it and ways of measuring it (depending on whether it is explored from the client or provider perspective, whether the focus is on technical information imparted during the exchange, the rapport between client and provider, the service delivery environment and its privacy and availability of other information such as leaflets, client recall of specific messages about side effects versus their general recall as to whether more than one method was discussed, use of mystery clients, direct observations, client-exit interviews immediately after the service, household surveys that occur much after the service exchange, assessments of provider biases and knowledge, etc.). It is crucial to acknowledge this and describe this paper’s definition of FP counseling quality in that context. The current draft seems to assume that FP counseling quality and information exchange are the same thing when in fact, I think, the latter is an aspect of the former.

Then proceed to explain the choice to focus specifically on the “information exchange”, why this is novel and appropriate. Some useful information for this is already in the paper, but it is not clear what dimension of quality and family planning quality that they measured and report on.

“Studies on family planning counseling service in Ethiopia showed that 40 the coverage is below 40%” – What does this mean exactly?

- Percentage of women who received high quality family planning counseling service is below 40%. The sentence updated in the revised version of the manuscript.

Lines 42-49, make clear the type of data you are talking about when you are citing the literature. Were these household surveys in which women were asked to recall information about the last time they sought FP services? Or were they exit-interviews conducted immediately after women were discharged from care? Did the surveys assess quality from the client or provider perspective (i.e., who did they enroll)?

- It is a household surveys in which women were asked to recall information about the last time they sought family planning counseling service. So, the survey assesses the quality from client perspective.

Materials and Methods

“The PMA surveys are repeated cross-sectional surveys based on a multistage stratified cluster sampling design.” Who do they interview and where? What is the operational definition of the enumeration area” and what is the operational definition of “population proportion to size” for these surveys? What does the term “multistage stratified cluster sampling design” mean for this study specifically (define the stages, define the cluster and strata). Did the sampling frame ever change during the seven years of survey implementation and if so how often? I prefer to see this information in the paper and not refer to another paper. I am familiar enough with PMA 2020 to know that it is well done, but I find this information rather crucial to understand the rest of the manuscript so I think you should provide it.

- Thanks for the feedback, and updated accordingly.

You refer to “the outcome variable” in line 70 but at this juncture I do not think that you have formally defined it yet.

- Addressed.

Variables: I find this section extremely difficult to understand because I do not know if the survey respondents were women of reproductive age enrolled in households, actual clients enrolled immediately after discharge from the service, providers that delivered the service and report of their own performance. Since I am familiar with PMA 2020 I assume you mean WRA enrolled at the household, so I can make it through the manuscript. But describe the sample clearly and how they are distinguished from the underlying population. Presumably your sample is a sub-sample of women from the surveys that had received FP services within some time frame before the survey, or are they simply reporting on the last time they ever sought FP services? Are they all FP users or acceptors or just anyone that received any counseling? Clarify the range of the duration representing the time between when they index counseling interaction took place and the time of the interview.

- Thanks for this comment. In the revised manuscript, study samples were clearly described.

Table 1: The third age category seems quite broad. I suggest breaking it down to 25-34, and 35+. What about permanent methods. Are you able to find out if they are new or return method users? Media exposure: do you mean exposure to media before or after the service episode. This may be a minor point, but consider whether it is relevant that the client-level factors represent the status of the client at the time of the interview and not when they obtained family planning.

- We categorized age as adolescent, young and adult for interpretation purpose. In the revised version we consider your comments, and recategorized age as: 15-19, 20-24, 25-34 and 35-49.

- Media exposure before the service episode.

I encourage the author to map out a causal pathway model between each proposed “exposure” and the “outcome” (quality of information exchange) taking into consideration the possibility of the exposure occurring after the dependent variable is mentioned, the theoretical linkages between the exposures and the quality of FP counseling, and possible recall bias. Its not necessary to include this in the paper, but the author may find it beneficial to do so and it is up to her/him.

In this vein, some issues to consider:

- Some of the client level factors may predispose clients to be more or less assertive during the interaction with provider (e.g., ask questions, request clarifications, inquire about different methods they have heard of). In this case, would the outcome reflect variation in the quality of counseling or rather the common sense that information exchange is better when clients ask questions.

- Some of the client level factors predispose some clients to poorer recall of counseling information than others.

- I am concerned about “FP discussion” as a exploratory variable (presumably you mean independent variable, possibly a predictor of interest)? How is an “FP discussion” different from the information exchange whose rating is treated as the outcome? Is this not problematic for your model?

- Thanks for raising this issue. We acknowledge your comment, and in the revised manuscript, we drop this variable from our analysis/model.

- I would include type of facility since I know they are many levels of care in the Ethiopian healthcare system (e.g., health post, health center, hospital, pharmacy, community based distributor).

- Thanks again. Yes! We did it now.

- Also it makes a difference if the woman is receiving a re-injection or refilling a prescription of oral contraceptives (she may not need counseling since she has been on the method for some time anyway, and I imagine you’ll find that counseling quality is lower for short-term methods for that reason).

- Yes, that is true.

- Can you include cadre type in the model (e.g. health extension worker, nurse, physician)?

- The point you raised is interesting, but we don’t have this data in this survey.

- The satisfaction variable seems problematic to me for the same reasons that I pointed out with respect to the “FP discussion” – it seems like a dimension of FP counseling quality in the same way that the information exchange is and I do not see how one can sensibly include in both sides of the equation.

- Thanks for raising this issue. We acknowledge your comment, and drop this variable from our analysis/model.

- Some one that received information on contraceptive method from a provider, especially a user of a method, may be more likely to report media exposure that occurred after the service because they were predisposed to being interested in family planning.

- Similarly exposure to media before the service may prompt women to seek family planning services and ask questions during the exchange with the provider. Thus, (1) there may be the problem of reverse causality, and (2) does this really tell us anything about the quality of counseling or rather the effect of media exposure on recall, health behaviors and assertiveness of client during exchanges with provider?

- Having media exposure on family planning may help the client to seek more information on the three key question (MII defined) from the provider which ultimately improve the quality of counseling.

- As you think through the causal pathway model, consider other factors such as whether the facility had commodities and supplies when the client sought FP services, whether the provider had ever received FP training and when.

- Thanks for the feedback. In the future, we will consider to collect such data in the upcoming surveys.

Statistcal analysis

Describe in greater detail how the “data analysis methods considered sampling weights for generating unbiased population estimates” and why the authors believed that this was necessary in the multi-level model with covariates? How did you incorporate the design-based approach to address unequal sampling probabilities in the multilevel model? Reading farther into the manuscript I see that you (might) have gone through the weighted procedures to obtain the descriptive results, but this also needs to be explained more clearly. Make very clear what were the different components of your analysis and the specific methods you implemented for each respectively. Generally, the lack of clarity in this manuscript has made reviewing this article time consuming and difficult. The authors should have their work reviewed by peers that can give this feedback before submitting to journals.

Include a statement that in as complete and plainspoken terms as possible gives the interpretation of the relationship between your independent and dependent variables that your analysis seeks and how the POM model ascertains this. The application of the POM to longitudinal data with repeat cross-sections is novel and interesting, but complex and readers should be guided in understanding how to interpret that aspect of the model as well. But, a reader that has never done ordered logistic regression but is otherwise conversant in regression will not be able to understand what you are saying. If I had not recently done a similar analysis myself I would be very confused.

Did you apply the Brant test to both models or just model 2?

- In both models. But now, since we drop FP discussion from our model, and re-categorize method source as Hospital, health-center, health-post, pharmacy and other, the proportional odds assumption satisfied.

-

Results

Table 3 is dizzying. I suggest you innovate with plots to convey the most important trends through a series of carefully tailored visualization, and elucidate these with parsimonious commentary and include this table as a Appendix only.

- We considered your suggestion and updated accordingly.

I think you need to reconsider the findings in light of some of the comments I made above about recall biases, counseling for STM v. LARC, reverse causation in association between media exposure and the outcome and what this tells us about counseling quality, the appropriateness of “FP discussion” and “satisfaction” as predictors, and absent data on the readiness for FP service delivery (available methods in stock, trained providers). In my opinion, the findings section needs to be strengthened quite appreciably to withstand that critique.

Your explanation of the model results does not reflect the ordered nature of the response variable. How do you interpret the OR in terms of the predictor response association reported by the POM?

- Thanks for pointing this. Updated.

Discussion

I suspect that this section may change significantly upon reflection of my comments above. The trend in the quality of the information exchange during FP counseling is noteworthy and merits elucidation in a published paper, but it seems that this analysis has missed the mark.

There is no question that provider bias is at play, but is it really fair to point to this as the driver of the national trend of declining FP counseling quality when the analysis does not acknowledge anything about the providers’ background, training, working environment, access to commodities and appropriate supervision?

Are LARC available in the private sector and if so is this restricted to particulate social franchises? Or are they mostly available in the public sector. It makes sense that LARC provision is associated with associated with more information provision since they entail removal services (unlike oral and injectable methods) and often times these occur in public sector settings. This does not mean that public sector providers are better counselors but the specialize in services that oblige them to share more information.

- As you suggested above, in the revised version of the manuscript, rather than public Vs private, we categorize the method source by facility type.

The issue of recall bias needs to be addressed squarely acknowledging the possibility that respondents may conflate the information they receive about FP methods from their provider with FP information that they heard from another source either before or after they had received FP services.

- Yes, that is true and we mentioned this fact under the limitation Section of the manuscript.

I think a more interesting analysis would focus on the individual characteristics of the provider (training, age, job title, sex, etc.), working environment (facility type, access to commodities, supervision, etc.), and macro-context (region, etc.). Pinpointing the quality of information that comes from provider to the client on characteristics of the client raises to the fore issues of provider bias that we already know about but cannot disentangle from the possibility that women with certain characteristics are more likely to have better recall than others.

- We agree with your suggestions, but in our survey considered for this analysis, we don’t have data about the individual characteristics of the provider linked with each client.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

17 Jan 2022

26 Mar 2022

Thank you for the opportunity to respond to reviewer comments. We thank the reviewer for a thorough secondary review.

Additional Editor Comments:

This is an important analysis as I expressed in the previous iterations. The manuscript has improved and the authors addressed most of my comments/questions.

My previous question remaining or with unsatisfactory response:

1. I did ask about the missingness before. Clarifying, what is the response rate of these PMA surveys? What was done if there were participants with missing data among the variables included for this analysis? This could be placed between the current lines 77 to 83.

Thank you to the reviewer for catching this and we apologize for the oversight in the previous version. We have added the text below on page 3 in the revised clean manuscript.

- In PMA surveys considered in this analysis, the response rate is high across different survey years which ranges from 95.6% to 99.2%. The analysis was done for women who completed the survey. If a woman replied “Don’t know or No response”, these values were recoded as missing.

New few issues:

1. The reviewer below, points out the need for some introduction of the concept family planning (FP) service quality and how that end be measured as just 3 questions.

We have expanded our discussion of the Method Information Index in the manuscript. While we agree that three questions are unlikely to capture the full complexity of the concept of quality, the Method Information Index is a well-established indicator in the field of family planning and generally accepted as the current standard for measuring quality. We have included additional information on the creation and validation of the indicator in the Background section and in the abstract.

“Counseling plays a key role in enhancing family planning services, providing contraception related information and supporting family planning and fertility goals for women of childbearing age [4]. The Method Information Index (MII) is the current standard proposed by the Track 20 initiative to monitor quality of family planning counseling [5, 6]. The simple index uses three questions -whether the respondent received counseling on multiple contraceptive methods, whether she was told about side-effects, and if so, whether she was told what to do about side-effects that are available in the majority of population-based surveys, including the Demographic and Health Survey (the DHS) and Performance Monitoring for Action (PMA) [7]. Adaptations have been proposed to the MII, including a fourth question assessing whether women were told they could switch to another method. This question was

added in PMA 2019 Survey, however, and for trend purposes is not included in the current analysis.”

2. Line21: What is this coverage? Is this high quality of FP coverage?

- Thank you. Yes, this has been updated.

3. Line 63: the SNNP abbreviation is used for the first time here. Can you write fully what it means?

Yes done. SNNP stands for Southern Nations, Nationalities, and Peoples'

4. Just for clarification: was the year of survey included in the model as a categorical variable. Correct? Can you clarify that in table 1.

- Yes, and updated accordingly.

5. Results related to table 2. Can you clarify how the weights of the two surveys were combined for 2014 analysis?

- Since the two surveys in 2014 conducted at the beginning and end of 2014, we recalculate the weights using the projected women in reproductive age (15-49) population in 2013.5 and 2014.5 which is 22326082 and 23198829, respectively. See the details below (under #7) how it is calculated.

6. Lines 159: what were the conclusions of the Brant tests? Was the PO assumption kept across all variables?

- Yes, we have included this on page 6.

7. The paragraph 171 to 174 states that all analysis used the sample weights. This leads to 2 questions:

a. How multiple survey weights were combined for this analysis?

- Following the recommendations by Ruilin Ren who is a senior sampling statistician at ICF, we recalculate the survey weights to compute the percentage of women who got high quality family planning counseling service as follows.

denormalizedWeight=FQweight×(total females age 15-49 in the country at the time of the survey)/(number of women age 15-49 interviewed in the survey)

- Normalizeweight= denormalizedWeight/sum(denormalizedWeight)

- Where, FQweight represents the computed weight in each survey.

Year 2014 2015 2016 2017 2018 2019

Projected population of women 15-49

23,198,829

24,084,767

24,949,201

25,832,132

26,725,476 27,620,498

b. It is very tricky to do random-effects models with survey weights. How the results would differ if you did not use weights. This could be placed in the supplements if indeed you did a random-effects model with sample weights.

- Thank you for the comment. As mentioned in line #140 and line #187 weights were used during the estimation of proportions and their respective confidence intervals. We did not consider sample weights when we fit the random-effects model.

8. The limitations should be part of the discussion as the reviewer points out below. Somewhere the last 3 paragraphs would be OK.

- Thank you. As most family planning research tends to include the limitations discussion near the end of the discussion section, we have inserted as the penultimate paragraph.

-

9. Thank you for the good plots. Just few things:

a. All of them should have the vertical y-axis start at zero. In ggplot2 there is something like expand_limits(y = 0) to sort that.

b. Make all of them to have the same maximum. It is easy to misunderstand the plots as they have different scales.

- Thank you for the helpful comments. In the revised version, the figure axis was updated accordingly.

Reviewers' comments:

Abstract

In methods section it is written “women aged 1549” but it should say “15 to 49”.

- Updated accordingly.

In results section what is meant by “the percentage of women that received high family planning counseling service…” Do you mean high quality FP services?

- Yes, thank you. This has been updated throughout.

I am not sure if I use the term “determinants” to characterize the factors that you found were associated with FP counseling quality. For example, I think media is a contextual factor that may be associated with experiencing better counseling quality, but I question whether it truly determines the level of counseling quality.

- We thank the reviewer for their suggestion, and have replaced “determinant factors” with “factors”.

Even though this may cause the authors to exceed the word limit for the abstract, I think the abstract would be strengthened if it briefly stated how the authors define FP counseling quality.

- Thank you. We have added this to the introduction section of the abstract.

Background

The new paragraph that points out that in a study on MII across 25 countries “residence, education, household wealth and method type were generally found to be associated with receiving higher quality care” … To be helpful, please specify residence where and what method types were associated with higher quality care.

- Thank you. We have clarified in the introduction section that urban residence was associated with receipt of higher quality care and included additional information on method type. The updated language on page 2 is copied below.

Despite substantial global efforts to improve conceptualization, measurement, and delivery of high-quality family planning services, studies consistently find that receipt of high-quality services is low. In a study exploring the levels and trends of the MII across 25 countries, urban residence, education, household wealth, and method type were generally found to be associated with receiving higher quality care [?]. In all countries combined, the median MII values show that the contraceptive information received by women varied by method type and was highest for implant users and lowest among women relying on sterilization.

In the same paragraph, when you say “individual level factors associated with receiving high quality care” it would be useful to clarify that you mean client-level (as opposed to individual provider-level).

- Thank you for this useful comment. We have rephrased the term “individual level” factors to “client-level” factors throughout the document.

Materials and Methods

Are you able to account for whether or not the survey participants were new or returning users? This matters since it is possible that returning users had already received the counseling information in recent FP visits at their nearby facility.

- Thank you for this question. The PMA surveys asked about the information that was received at the visit at which the client first received their particular contraceptive method. This has been clarified in the manuscript and the implications for the potential for recall bias has been included in the limitations section. We note that the PMA questions use the same question wording as the DHS when collecting information on the MII and thus, while this bias does exist, it is consistent with previous analyses.

-

To be helpful, specify what methods you define as long-acting and short-acting.

Thank you. This has been clarified in the methods section. Specifically, we define long-acting methods as female sterilization, implant and IUD and short-acting methods as the injectable, pill, emergency contraception, male condom, the Standard Days Method, and lactational amehorrhea method (LAM).

Results

“More than 70% of women in all years got received their contraceptive methods…” Check for grammar.

- Thank you, this has been updated.

Line 258 on page 8: you mention a decline in the proportion of survey participants that report good counseling services, but you say it is a decline in coverage. Elsewhere you call it percentage of women received high quality FP counseling. I question whether coverage is the best term for this. I think it is a measure of the prevalence. Please be consistent in how you refer to the measure.

- Thank you, we have updated throughout.

Line 262 on page 8 references Table?? – please correct this.

Updated.

Lines 255-267 page 9 – please check to see if you have used the correct tense.

Figure 2 label- “trends in getting high family planning counseling…” should it not say high quality family planning?

- Thank you. We have performed a thorough check of grammar throughout.

I understand the purpose of Table 4, but it is incredibly confusing. I suggest a table with a row for every year in the time series and columns for year, % change between successive years including the 95% CI in parentheses and cumulative % change between current and start year with 95% CI in parentheses.

- Thank you for the helpful suggestion. The table has been updated based on the following format.

Table X: Summary of observed percentage changes in high quality family planning counseling service across survey-years.

Reference/ base year Survey year

2015 2016 2017 2018 2019

2014 39.9(19.57,60.25) 16.6(-3.03, 36.2) 14.3(-8.52, 37.2) 5.1(-16.1, 26.3) -57.3(-67.11, -47.48)

2015 -16.7(-28.10, -5.27) -18.3(-34.7, -1.9) -24(-40.1, 9.7) -69.47(-76.5,-62.4)

2016 -2.0(22.02,18.17) -9.84(-28.44,8.7) -63.4(-71.9,-54.8)

2017 -8.1(-21.84,4.69) -62.65(-71.05,-54.25)

2018 -59.4(-68.56,-50.18)

Lines 278-286- there seems to be some repetition here… please check and revise accordingly.

- Thank you, this has been updated.

Thanks for explaining the interpretation of Models A and B. After providing a thorough explanation of the interpretation for one of the explanatory variables, I do not think you need to repeat the same language over again for the other variables. But it is up to you.

-Thanks. This comment taken into account and we dropped the interpretation of some variables.

Discussion

It would be interesting if the authors could elaborate further on what might be the drivers of the decline in the outcome variable over the span of the time series.

- Thank you. We have added the following statement.

“Except in Addis Ababa, in all other regions, high-quality family planning counseling peaked in 2015 and declined since then in all other regions. The observed relatively better family planning counseling service in 2015 may be due to the fact the Ethiopian health sector transformation plan started in this year [31]. As compared with other

regions, high quality counseling service was low in Amhara region and sharply declined since 2017 (Figure 2). Nationally, our finding on the trend analysis in the percentage of women receiving high quality family planning counseling service declined overtime. This aligns with Hrusa's finding using PMA data [12].”

I think it is good that in the discussion you mention that the association you identified between media exposure and the outcome is likely due to the fact that the former enhances knowledge and may prompt clients to ask more questions. In link 382 of page 15 you say this supports women in receiving high quality services, but I think a better way of phrasing it is that media exposure enables a more informative client-provider interaction because it’s enhances client knowledge and empowers them to more inquisitive in the FP visit.

- Thank you, we have updated our language to better reflect the suggestion.

Can you establish whether women’s reports of FP media exposure occurred before their reported FP visits in each survey? If not, then you should acknowledge that since women that received FP services may be more likely to seek out or recall FP-related media exposure (e.g., the visit peaked their interest, women that listen to educational media programs are more educated and therefore also more likely to use FP and ask questions during the visit). In other words, the direction of the association between media exposure and FP quality cannot be determined by this analysis, which limits the interpretation.

The abstract mentions that seeking FP counseling from pharmacies was associated with lower levels of FP counseling quality. This is an interesting finding and I think the authors should elaborate on this further in the discussion.

Thank you. In the discussion Section, we have added the following statement.

“Women who received care from pharmacies had significantly lower odds of receiveing high-quality family planning counseling than women who received care at either a health post or health center. In addition to differences in service modality and training of service providers, this likely reflects differences in characteristics of users and method choice; other studies have shown that women who use pharmacies tend to live in urban areas, be younger, and rely on short-term methods [38, 39].”

In the discussion, you mention that the odds of receiving high quality FP counseling are higher among women that obtain the service at public facilities compared to those that sought it at private facilities and refer to table 5 (lines 393-397 on page 16). But Table 5 does not include columns that refer to public vs. private facilities. Table 5 does give parameter estimates for health post, health center, pharmacy and other facility (ref hospital). This finding could be elaborated on in the discussion, with some clarification on the difference in the Ethiopian health system between health posts and health centers.

- Thank you. We clarify that the health posts, health centers, and hospitals included are all public health facilities, while pharmacies are all considered private facilities. We have provided additional information on page 12.

It is interesting that the outcomes vary so appreciably by region and if the authors have valuable insight on why this then please include that in the discussion.

-Added in the second paragraph of the discussion Section.

Lines 357-367 indicate that the analysis has implications regarding the need for youth friendly services, even though you found that age was not associated with FP counseling quality in the fully adjusted model. It seems that including parity and age in the same model washes out an association between age and the outcome, which is not surprising. I do not think the findings give a useful interpretation related to provider biases and clients’ age, and you should revise or remove this section from the discussion. Lines 419-430 are similarly confusing. Line 419 says age is not significantly associated with the outcome in the fully adjusted model, contrary to other studies and then you proceed to reflect on the crude OR in Model A. If after adjusting for parity, you find there is no age effect, then you might consider a different interpretation. For example, as women become more experienced with childbearing their knowledge and confidence during interactions with SRH providers increase and they are able to ask for and recall more information during their FP visits. I concede that women with little or no childbearing experience are also young, but I think you need to frame the discussion on age/parity and the outcome better and in a manner more aligned with Model B findings.

- Thanks for your fruitful comments. Accordingly updated.

Limitations

Unless it is a formatting requirement of the journal, I think the limitations sections should be included in the discussion section (personally, I think it is best at the beginning of the discussion). I do not think it is a good way to end the article.

- Thank you for the suggestion. We have included limitations in the discussion section, but have chosen to place them in the penultimate paragraph of the paper, as we found this was in keeping with the majority of family planning manuscripts included in our references.

I think the final sentence of the limitations should be expanded to state, specifically, that the analysis lacks really critical supply-side variables, such as availability of methods at facilities, availability of trained staff, job aids, information leaflets for clients, etc. Also, the analysis doesn’t discuss whether survey participants were new or returning users, whether they had ever practiced FP in their lives, etc.

- Thank you. In the last line of the limitation paragraph, we included a statement which addresses the lack facility-level variables in the analysis and encourages further research on addressing this shortcoming.

-

Conclusion and recommendation

The conclusions right now seem rather generic and disconnected from specific lessons from the analysis. This is understandable but the authors need to reflect on what they can credibly conclude and give guidance on based on their findings and report this in a more focused manner.

The study seems to conclude that having more formal education, access to FP information from media and more experience with SRH care seeking (based on higher parity levels) is associated with recalling more FP counseling information. However, we also know from the analysis that counseling tends to be better at health posts and health centers compared to hospitals and is particularly poor at pharmacies, that LARC provision is association with better counseling, that some regions do appreciably better than others and that overall, counseling quality has declined over time.

- Thank you. We included more specific recommendations particularly highlighting the importance of providing additional training to pharmacists and related staff on contraceptive methods.

Though I am skeptical about this analysis because it does not include vital supply-side variable that affect counseling quality, I think the findings at least point in the direction of some supply- and demand-side actions that the MOH and partners should consider to address the problem of poor counseling quality.

A methodological recommendation to consider is incorporated more supply side measures (that PMA has made available from its extensive work conducting facilities assessments, I think) into future analysis on FP service quality.

Thank you. We have added this recommendation in the limitations section.

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

20 Apr 2022

Trend and Determinants of Quality of Family Planning Counseling in Ethiopia: Evidence from repeated PMA cross-sectional surveys, (2014-2019)

PONE-D-21-19351R2

Dear Dr. Ejigu,

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,

Orvalho Augusto, MD, MPH

Academic Editor

PLOS ONE

19 May 2022

PONE-D-21-19351R2

Trend and Determinants of Quality of Family Planning Counseling in Ethiopia: Evidence from repeated PMA cross-sectional surveys, (2014-2019)

Dear Dr. Ejigu:

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. Orvalho Augusto

Academic Editor

PLOS ONE