- Research article
- Open Access
- Open Peer Review
Reproductive health service utilization and associated factors among adolescents (15–19 years old) in Gondar town, Northwest Ethiopia
BMC Health Services Researchvolume 13, Article number: 294 (2013)
The utilization of reproductive health services is an important component in preventing adolescents from different sexual and reproductive health problems. It plays a vital role in safeguarding youth in Sub-Saharan African countries including Ethiopia, which accounts for a high proportion of the region’s new HIV infections as well as maternal and infant mortality ratios. Due to this, assessing adolescent reproductive health service utilization and associated factors has its own contribution in achieving the national Millennium Development Goals (MDG), especially goals 4 to 6.
A community based cross-sectional study was conducted from April 5–19, 2012, in 4 randomly selected administrative areas of Gondar town. A total of 1290 adolescents aged 15–19 were interviewed using a pre-tested and structured questionnaire. Data were entered in to the EPI INFO version 3.5.3 statistical software and analyzed using an adapted SPSS version 20 software package. Logistic regression was done to identify possible factors associated with family planning (FP), and voluntary counseling and testing (VCT) service utilization.
Out of the total participants, 79.5% and 72.2% utilized FP and VCT services, respectively. In addition, among sexually experienced adolescents, 68.1% and 88.4% utilized contraceptive methods and VCT service during their first sexual encounter, respectively. Educational status, discussion with family/relatives, peer groups, sexual partners and teachers were significantly associated with FP service utilization. Also, adolescents who had a romantic sexual relationship, and those whose last sexual relationship was long-term, were about 6.5 times (Adjusted Odds Ratio [AOR] = 6.5, 95% CI: 1.23, 34.59), and about 3 times (AOR = 3, 95% CI: 1.02, 8.24) more likely to utilize FP services than adolescents who had no romantic relationship or long-term sexual relationship, respectively. In addition, the variables significantly associated with VCT service utilization were: participants who had secondary education and above, schooling attendance, co- residence with both parents, parental communication, discussion of services with peer groups, health workers, and perception of a risk of HIV/AIDS.
The majority of the adolescents were utilizing FP and VCT service in Northwest Ethiopia. But among the sexually experienced adolescents, utilization of FP at first sexual intercourse and VCT service were found to be low. Educational status, schooling attendance, discussion of services, type of sexual relationship and perception of risk were important factors affecting the utilization of FP and VCT services. Building life skill, facilitating parent to child communication, establishing and strengthening of youth centers and school reproductive health clubs are important steps to improve adolescents’ reproductive health (RH) service utilization.
Reproductive health care is defined as the constellation of methods, techniques and services that contribute to reproductive health and well-being by preventing and solving sexual health problems . Globally, the adolescent population is estimated to be 1.25 billion . Among these, 513 million are between 15–19 years old , and 85% of the total adolescents are living in developing countries . These adolescents are most vulnerable to a range of reproductive health problems, such as teenage pregnancy and childbearing, unsafe abortion and sexually transmitted infections (STI), including HIV .
The World Health Organization (WHO) estimates that in Africa, 60% of all new HIV- infections occur in adolescents who are 15–19 years old . In developing countries, there are about 12.8 million births by adolescents aged between 15–19 years, and a large proportion of these pregnancies are unplanned . Among married women who are 15–19 years old, only 17 percent practice family planning methods currently, and among unmarried sexually active adolescents, the use of contraception is believed to be even lower . A WHO report in Algeria, Bangladesh, Ethiopia, Indonesia and Nigeria showed that the risk of dying from complications related to pregnancy or childbirth is two times higher for those aged 15–19 than for women in their mid-twenties .
According to the 2011 Ethiopian Demographic and Health Survey, contraceptive use among currently married women of 15–19 years of age was only 23%, with 0% utilization of permanent methods, 1.6% and 2.5% utilization of implants and IUD, respectively. Contraceptive use was lower when compared with other age groups. Among sexually active youth aged 15–19 years, women and men who were tested for HIV test were only 24% and 27% respectively . Therefore, assessing factors affecting reproductive health service utilization within this age category, especially FP and VCT services, are very important to improve adolescent reproductive health service utilization and thereby reduce the burden of adolescent disease and disabilities associated with reproductive health.
Study design and set up
A community based cross-sectional study was conducted in Gondar town, Northwest Ethiopia from April 5–19, 2012. Gondar town is found in North Gondar Zone of the Amhara Regional State, 750 km Northwest of Addis Ababa. According to the 2007 Ethiopian census report, Gondar had a total population of 206,987, and adolescents aged 15–19 years were estimated to be 12% (25,128) of the total population . The town is divided into 12 administrative areas. The participants of this study were adolescents 15–19 years old who had lived in the area for at least six months.
Sample size and sampling procedure
The sample size was determined by using the single population proportion formula with the following assumptions: 17.6% proportion , 95% confidence level, 3% margin of error and design effect of 2. Then five percent was added for the expected non-response, making the final sample size 1300. From the 12 administrative areas of Gondar town, 4 were randomly selected. The sample allocated was proportional to the household size of each administrative area. The first household from each administrative area was identified using the lottery method, and then the systematic random sampling technique was applied to identify the next household to be included. Adolescents who were found in the selected households were interviewed. In the case of more than one eligible participant in the household, the lottery method was used to select one. In households in which adolescents were not at home, but it was known that there were eligible adolescents for the study, the interviewers revisited the household at three different time intervals, and when interviewers failed to meet the adolescent, the household was excluded from the survey and replaced by the next household in a clockwise direction.
Data collection tool and procedure
A structured questionnaire was developed and administered to the participants. The questionnaire and consent documents were first developed in English, then translated into Amharic, and finally retranslated into English by another translator to check consistency. Before the actual work, data collectors and supervisors were given intensive training for two days about the aim of the study, procedures and data collection techniques by going through the questionnaire question by question. After the training, interviewers pre-tested the questionnaire on 78 (6%) people living in an administrative area not selected for the study.
The following operational definitions were used: Reproductive health services particularly considered in this study were FP and VCT services. FP service utilization was defined as ever use of any modern contraceptives in life time, and VCT service utilization referred to ever utilization of VCT service for HIV testing. Duration of sexual relationship was defined as number of months from first sexual intercourse to the last sexual intercourse. Type of sexual relationship reflects whether adolescents ever had a romantic relationship in their sexual experience or not.
Data processing and analysis
Data were checked for completeness and entered into EPI INFO version 3.5.3 statistical software and then exported to SPSS version 20 for further analysis. Multiple logistic regression was used to identify variables independently associated with service utilization. The strength of association was interpreted using the adjusted odds ratio and 95% CI. The criterion for statistical significance was set at a p value of 0.05.
Ethical clearance was obtained from the Institution Review Board of the Institute of Public Health, the University of Gondar. Formal letters of cooperation were written to different kebele administrators. Oral consent was obtained either from parents or surrogate and participants. Moreover, the adolescents were assured that neither the interviewer nor their parents would have access to their responses. Confidentiality was assured by using anonymous questions and by conducting the interviews privately throughout.
Socio-demographic characteristics of study the participants
Out of 1300 study samples, 1290 responded to our interview making the response rate 99.23%. About half of them (50.4%) were females and 1209 (93.7%) were single. The mean age of the participants was 16.96 years with a standard deviation of 1.4 years. More than three-fourths of the participants (85.5%) were enrolled in school and 647 (50.2%) had at least secondary education. More than one third (44.2%) of the participants’ mothers’ age was between 40–49 years and 699(55.2) of them had no formal education. Around 61.7% were living with both parents (Table 1).
Sexual history of the adolescents
Out of the total participants, 445 (34.5%) have had sexual partners, and 303 (23.5%) have had sexual intercourse. Out of those who had sexual partners, the majority (84.5%) had one sexual partner. Among those who had sexual intercourse, 294 (97%) had sexual intercourse in the past 12 months and 223 (73.6%) had sexual intercourse more than once with the same sexual partner. Out of those who had sexual intercourse, 269 (88.8%) had a romantic sexual relationship, and 218 (71.9%) stayed in their last sexual relationship for more than six months, with a median duration of 12 months (Table 2).
Family planning service utilization
About 1205 (93.4%) of the participants had awareness about family planning services. One thousand sixteen (82%) of the participants had discussed the service. The majority of the participants, 899 (88.5%), discussed the service with their peers and 667 (65.6%) with their teachers.
Out of those who had sexual intercourse, 241 (79.5%) utilized family planning services. One hundred sixty-four (68.1%) of these used contraceptive methods at their first sexual intercourse, and 234 (97.1%) used contraception at their last intercourse. The most commonly used contraceptive method was the condom, 112 (46.5%), followed by injectables, 103 (42.73%). More users went to health centers, 127 (52.7%), while others, 103 (42.7%), preferred shops for the service. The major reasons for not using family planning services for 57 (87.1%) of the participants was fear of being detected. Others, 37 (59.7%) reported that they disliked the judgmental attitude of health workers.
Voluntary counseling and testing service utilization
Out of the total participants, 1265 (98%) had heard about the VCT service. The majority of the participants (89%) had discussed the service with someone. Peer groups/friends, 1049 (91.4%), followed by teachers, 771 (67.2%), were the main people with whom the participants of the study discussed VCT services. One fourth of the participants (25%) perceived themselves as being at risk to contracting HIV/AIDS. Out of all adolescents, 932 (72.2%) utilized the VCT service, and out of the sexually experienced, 268 (88.4%) utilized the service. More than half (55.6%) were females, and 608 (65.3%) obtained the service from health centers and 220 (23.61%) from schools. Partner or self trust, 326 (91.1%), and embarrassment, 226 (63.1%), were the main reasons for participants for not using the VCT service.
Factors associated with family panning service utilization
Based on the bivariate analysis, the factors found to be significantly associated with family planning service utilization were age, schooling attendance, educational status, mother’s educational status, co- residence with both parents, parental monitoring, parental communication, discussion on family planning services, number of sexual partners, amount of sexual experience, having a romantic sexual relationship and duration of last sexual relationship.
Out of variables which were entered to multiple logistic regression, participants’ educational status, maternal educational status, discussion on FP use with family/relatives, peer group, sexual partners and teachers, having romantic and long term sexual relationships were found to be significantly associated with family planning service utilization.
Adolescents with secondary education and above were about 9 times more likely to utilize FP service as compared to those with no formal education (AOR = 9, 95% CI: 1.45, 54.14). Similarly, maternal education was found to have an association with the utilization of family planning services (AOR = 6.4, 95% CI: 1.39, 29.72).
Having romantic sexual relationships and a long duration of the last sexual relationship were also found to be strong predictors of FP service utilization. Adolescents who had romantic sexual relationships were about 6.5 times (AOR = 6.5, 95% CI: 1.23, 34.59) more likely to utilize family planning services than those who had not had romantic sexual relationships. In addition, adolescents who had a long duration of the last sexual relationship were about 3 times (AOR = 3, 95% CI: 1.02, 8.24) more likely to utilize the service than those had a short duration of last sexual relationship.
Discussion with family/relatives, peer group/friends, sexual partners and teachers on family planning were also the other factors significantly and independently associated with the utilization of the service (Table 3).
Factors associated with voluntary counseling and testing service utilization
On bivariate analysis, the factors that were found to be significantly associated with VCT service utilization were sex, age, schooling attendance, educational status, mother’s educational status, co-residence with both parents, parental monitoring, parental communication, discussion of VCT services, having ever had a sexual partner, having ever had a sexual experience and perception of risk towards HIV/AIDS.
Out of variables which were entered to multiple logistic regression, sex, schooling attendance, educational status, co-residence with both parents, parental communication, discussion of VCT services with peer group and health workers, sexual experience and perception of risk towards HIV/AIDS were found to be significantly associated with VCT service utilization.
Female adolescents were about 2.6 times (AOR = 2.6, 95% CI: 1.79, 3.80) more likely to utilize VCT service than the males. The odds of having VCT was about 2 times (AOR = 2, 95% CI: 1.09, 3.41) higher for in school adolescents than out of school ones. Similarly, adolescents with secondary education and above were about 3 times more likely to use VCT as compared to those who have no formal education (AOR = 3, 95% CI: 1.11, 6.78).
Participants living with both parents were about 1.5 times more likely to use VCT services (AOR= 1.5, 95% CI: 1.02, 2.12). Adolescents who had had parental discussion on VCT services were 10times (AOR = 10, 95% CI: 6.09, 17.55) more likely to utilize the service as compared to those who had no parental communication. Discussion with peer groups/friends and health workers were the other variables associated with the utilization of VCT services.
Adolescents who had ever had sexual intercourse were about 4 times (AOR = 4, 95% CI: 2.09, 6.89) more likely to use VCT services than abstainers. Perceiving a risk towards HIV/AIDS was also found to be one of the predictors for the utilization of VCT. Adolescents who had a perception of risk towards HIV/AIDS were about 30 times more likely to utilize VCT services than those who had no perception of risk at all (AOR = 30, 95% CI: 10.65, 83.01) (Table 4).
The percentage of adolescents who attend health facilities for FP and HIV preventive services is one of the indicators of the immediate and long term RH needs of young people . This community-based study which assessed the patterns of adolescent FP, VCT service utilization and associated factors is important to see the progress towards achieving the national MDGs, especially goals 4 to 6.
The study revealed that utilization of FP services was 79%; this finding was higher than those of studies conducted in Jimma (17.6%), Eastern parts of Gojam (21%) and Ghana (49%) [10, 12, 13]. The possible reason for the difference could be that the Jimma study consisted of both sexually experienced & inexperienced adolescents, whereas this study strictly focused on sexually experienced ones. Differences in the composition of the study subjects might be the reason for the different findings of Gojjam and Gondar; in Gojjam the study was conducted among out of school adolescents who were married and illiterate.
Experimentation with sex is a natural and normal part of adolescence, but experimentation without protection is one of the indicators of risky sexual behavior. This study revealed that contraceptive use at first sexual intercourse was only 68%. This finding showed how much these adolescents were at risk to different sexual and reproductive health problems, like unwanted pregnancy, unsafe abortion, STI’s and HIV/AIDS.
This study found that utilization of VCT service was much higher than a study conducted in Wollisso Woreda which was 5.3%, and in rural Butagira at 6% [14, 15]. This is because the perception of risk of HIV/AIDS among sexually active adolescents was less than the results of this study. In addition, in the Wollisso area, there was low access to VCT service since there were only 3 facilities which provided VCT services. Moreover, lack of continuous supply of HIV testing re-agents and kits were the other problems which discouraged adolescents from utilizing VCT services. The difference in the findings of the Butagira study might be that the Butagira study was predominantly conducted among rural youth, in which the majority were out of school and had high levels of illiteracy. This in turn could be due to the fact that youth in rural areas have low access to information and VCT services.
In order to reach the UNAIDS goal of a 30 percent reduction in new HIV infections among young people by 2015, one of the measurable results is increasing the number of young people who know their status through counseling and testing services. Besides, bolstering these services and sustaining them among young people will be crucial in achieving “zero new HIV infections, zero discrimination and zero AIDS-related deaths”. However, this study revealed that the utilization of VCT among sexually experienced adolescents was only 88.4%.
In this study, the overall utilization of VCT services among males and females was 65% and 80%, respectively. These findings were higher than that of a study conducted in South Africa which was 9.7% for males and 10.9% for females . The South Africa study was conducted in the rural areas of Eastern Cape Province, where HIV rapid tests were available at a small minority (approximately 10%) of clinics. According to the 2002 facility survey, HIV/AIDS counseling was routinely available in the rural areas of South Africa, but access to testing procedures was less common.
A substantial number of studies identified that adolescents with secondary education and above were more likely to utilize FP and VCT service [15–20]. This study also supported the above claim. This can be explained by the fact that educated adolescents have an increased knowledge about the availability of the service, the benefit of preventive health care, and have a higher receptivity towards new health-related information and better communication with their sexual partners.
Educated mothers are more open to discuss FP issues with their children. They are also more flexible to deal with problems faced by their children regarding reproductive health service utilization. Secondary and above educational attainment of the mother was also one of the factors significantly associated with FP service utilization. This finding is in line with a study which was conducted in East Gojam . In addition, this research found out that discussion of the service with sexual partners, family/relatives, peer groups/friends and teachers had a significant association with FP service utilization. This can be justified by the fact that discussion of services with different categories of people allows adolescents to create more opportunities to exchange information, experiences, and build comprehensive knowledge about FP. It can also create opportunities to deal with adolescent problems associated with FP service utilization. Adolescents who have had romantic and lengthy sexual relationship were more likely to utilize FP service. This finding is consistent with studies conducted in the USA [21, 22]. This could be because adolescents who have had romantic and lengthy sexual relationships may have more time to plan and discuss methods of protection to use during sexual activity. They are also more likely to care about each other and try to alleviate problems they may be facing during sexual relationships.
Female adolescents were more likely to utilize VCT services as compared to male adolescents. This finding is different from that of a study conducted in South and North Nigeria . A possible explanation can be that the factors which help facilitate the utilization of VCT in this study (like perception of risk, having parental communication and discussion about the service with different categories of people) were found to be much lower among male participants than female ones.
School connectedness is one of the protective factors for adolescents in the ecological model of adolescent health. This study is in favor of the above association. Schooling status creates an opportunity to engage with different health promotion programs, like school based VCT service programs and school clubs which allow the adolescents to obtain more new information and knowledge related to the service.
Adolescents who were living with both parents were more likely to utilize VCT as compared to those who did not. This finding is against a study conducted in the UK which revealed that those who were living with single parents were more likely to utilize the service than those who were living with both parents . This might be because most of the adolescents in the UK have a higher level of parental monitoring than the subjects in this study. This in turn may hinder service utilization. Living with both parents increases the likelihood of communication and discussion with adolescents, as well as helping shape adolescent behavior.
In this study, discussions with parents, peer groups/friends and health workers were significantly associated with adolescents’ VCT service utilization. This finding is consistent with a study done in Zambia . This might be due to the fact that social relationships may influence young people’s decisions regarding HIV testing. Moreover, adolescents are more likely to engage in healthy behaviors when they feel connected to their family and peers. Since peer groups/friends are composed of individuals within similar age ranges, it also creates an opportunity to share information and experiences regarding VCT service.
Perceiving a risk towards HIV/AIDS was also found to be a strong predictor of utilization of the service. This finding is similar with studies conducted in Nigeria and Zambia. These studies revealed that those perceiving a risk towards HIV were more likely to go for VCT than those who believed they were not at risk of contracting HIV [22, 24]. This might be due to the fact that adolescents will not seek healthcare, like being tested for HIV, unless they view themselves as potentially vulnerable. Therefore, perceiving a risk towards HIV/AIDS is a cue to seeking healthcare, like knowing one’s own HIV status.
Generalization of the findings presented should be made with caution because of the following limitations: this study shared the limitations of other cross sectional studies, i.e. the difficulty of determining causal relationship between variables. Social desirability bias may have resulted in underreporting of family planning, voluntary counseling and testing utilization. Recall bias may also affect responses about events in some of the responses.
In conclusion, the majority of adolescents were utilizing FP and VCT services in Northwest Ethiopia. But among sexually experienced adolescents, utilization of FP services at first sexual intercourse and VCT services were found to be low. The major reasons for not using family planning services were fear or embarrassment, followed by judgmental attitudes of health workers. In addition, partner or self-trust and embarrassment to use services were the major reasons for not using VCT. Thus, it is suggested that there be an intensified effort to increase the health service utilization for RH by adolescents, as they are the prime victims of a range of RH problems. Building life skills, facilitating parent to child communication, establishing and strengthening of youth centers and school reproductive health clubs are important steps to improve adolescent reproductive health service utilization.
Roudi-Fahimi F, Ashford L: Sexual and Reproductive Health in the Middle East and North Africa: A Guide for reportors. 2008, Washington DC: Population Reference Bureau
WHO: Promoting Adolescent sexual and reproductive health through schools in low income countries; an information brief. 2008, Geneva: WHO
Kirby D: A proposed adolescent reproductive health intiative:Population Technical Assistance Project. 1994, Cairo: USAID, Office of Population
Belachew T, Abebe G, Nigussie S, H/Mariam A, Taye K, Yihdego D, Assefa D: Manual on Reproductive Health. 2003, Ethiopia: EPHTI
Ingwersen R: Adolescent reproductive and sexual health in the developing world. 2001, Canberra Australia: Development Studies Network
Nwobodo E: Reproductive health issues of the adolescent in Africa: Review and commentary. Niger J physiol Sci. 2002, 17: 1-8.
WHO: Adolescent pregnancy –Unmet needs and undone deeds: A review of the literature and programmes: Issues in Adolescent Health and Development. 2007, Geneva: WHO
Central Statistical Agency [Ethiopia] and ICF International: Ethiopia Demographic and Health Survey 2011. 2012, Addis Ababa, Ethiopia and Calverton, Maryland, USA: Central Statistical Agency and ICF International
Federal Democratic Republic of Ethiopia Population Census Commission: Summary and statistical report of the 2007 population and housing census: Population size by age and sex. 2008, Addis Abeba: Population Census Commission
Tegegn A, Gelaw Y: Adolescent Reproductive Health Services in Jimma city, Accessibility and Utilization. Ethiopian J Health sci. 2009, 19 (2): 91-102.
FMOH, UNFPA, UNAIDS, WHO: National adolescent and youth reproductive health strategy in Ethiopia. 2006, Addis Ababa, Ethiopia: FMoH
Seifu A, Fantahun M, Worku A: Reproductive health needs of out-of-school adolescents: A cross-sectional comparative study of rural and urban areas in northwest Ethiopia. Ethiopian J Health Developement. 2006, 20 (1): 10-17.
Glover EK, Bannerman A, Pence BW, Jones H, Robert M, Weiss E, Nerquaye-Tetteh J: Reviewed: Sexual Health Experiences of Adolescents in Three Ghanaian Towns. Int family plann perspect. 2003, 29 (1): 32-40. 10.2307/3180999.
Mekonnen Y, Demissie T: KAP study on HIV/AIDS among adolescents and assessment of available HIV/AIDS related services in Wolliso Woreda. 2005, Addis Ababa, Ethiopia: Save The Children-USA
Molla M, Berhane Y, Lindtjørn B: Perception of Ethiopian youth regarding their risk of HIV, A community-based study among youth in predominately rural Butajira. Ethiopian J Reproductive Health. 2009, 3 (3): 44-51.
Hutchinson PL, Mahlalela X: Utilization of Voluntary Counseling and Testing services in the Eastern Cape, South Africa. AIDS care. 2006, 18 (5): 446-455. 10.1080/09540120500213511.
Martin TC: Contraceptive use patterns among Spanish single youth. Eur J Contraception and Reproductive Health Care. 2005, 10 (4): 219-228. 10.1080/13625180500282379.
MacPhail C, Pettifor AE, Pascoe S, Rees HV: Contraception use and pregnancy among 15–24 year old South African women: a nationally representative cross sectional survey. J BMC Med. 2007, 5 (31): 1-8.
Katende C, Gupta N, Bessinger R: Facility-Level Reproductive Health Interventions and Contraceptive Use in Uganda. Int Fam Plan Perspect. 2003, 29 (3): 130-137. 10.2307/3181079.
Nwachukwu CE, Odimegwu C: Regional patterns and correlates of HIV voluntary counselling and testing among youths in Nigeria. Afr J reproductive health. 2011, 15 (2): 131-146.
Manlove J, Ryan S, Franzetta K: Patterns of Contraceptive Use Within Teenagers’ First Sexual Relationships in Washingtone DC. Perspect on sexual and reproductive health. 2003, 35 (6): 246-255. 10.1363/3524603.
Ford K, Sohn W, Lepkowski J: Characterstics of Adolescents sexual partners and their association with use of condoms and other contraceptive method in USA. Family plann perspect. 2001, 33 (3): 100-105. 10.2307/2673765.
Parkes A, Wight D, Henderson M: Teenagers’ use of sexual health services, Perceived need, knowledge and ability to access. J family plann reproductive health care. 2004, 30 (4): 217-224. 10.1783/0000000042177225.
Denison JA, McCauley AP, Dunnett-Dagg WA, Lungu N, Sweat MD: HIV testing among adolescents in Ndola, Zambia: How individual, relational and environmental factors relate to demand. AIDS Educ Prev. 2009, 21 (4): 314-324. 10.1521/aeap.2009.21.4.314.
Obermeyer CM, Osborn M: The utilization of testing and counseling for HIV: a review of the social and behavioral evidence. Am J Public Health. 2007, 97 (10): 1762-1774. 10.2105/AJPH.2006.096263.
The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6963/13/294/prepub
Our heartfelt thank goes to University of Gondar for financial support and all study participants who voluntarily participated in this study.
The authors declare that they have no competing interests.
SAF wrote the proposal, participated in data collection, analyzed the data and drafted the paper. ZBM, AFD and DNK approved the proposal with great revisions, participated in data analysis and revised subsequent drafts of the paper. All authors read and approved the final manuscript.