Kawakatsu et al.

BMC Pregnancy and Childbirth 2014, 14:265

http://www.biomedcentral.com/1471-2393/14/265

RESEARCH ARTICLE Open Access

Determinants of health facility utilization for

childbirth in rural western Kenya: cross-sectional
study
Yoshito Kawakatsu1,2*, Tomohiko Sugishita1, Kennedy Oruenjo3, Steve Wakhule3, Kennedy Kibosia3, Eric Were4
and Sumihisa Honda2

Abstract
Background: Skilled attendance at delivery is recognized as one of the most important factors in preventing
maternal death. However, more than 50% of births in Kenya still occur in non-institutional locations supported by
family members and/or traditional birth attendants (TBAs). To improve this situation, a study of the determinants of
facility delivery, including individual, family and community factors, was necessary to consider effective intervention
in Kenya.
Methods: This study was conducted to identify the factors which influence the place of delivery in rural western
Kenya, and to recommend ways to improve women’s access to skilled attendants at delivery. A community-based
cross-sectional survey was carried out from August to September 2011 in all 64 sub-locations which were covered
by community health workers (CHWs). An interviewer-administered questionnaire on seventeen comprehensive
variables was administered to 2,560 women who had children aged 12–24 months.
Results: The response rate was 79% (n = 2,026). Of the respondents, 48% of births occurred in a health facility and
52% in a non-institutional location. The significant determinants of facility delivery examined using multivariate
analysis were: maternal education level, maternal health knowledge, ANC visits, birth interval, economic status of
household, number of household members, household sanitation practices and traveling time to nearest health facility.
Conclusions: The results suggest that the involvement of TBAs to promote facility delivery is still one of the most
important strategies. Strengthening CHWs’ performance by focusing on a limited number of topics and clear
management guidance might also be an effective intervention. Stressing the importance of regular attendance at ANC
(at least four times) would be effective in enhancing motivation for a facility delivery. Based on our findings, those actions
to improve the facility delivery rate should focus more on pregnant women who have a low education level, poor health
knowledge and short pregnancy spacing. In addition, women with low economic status, a large number of family
members and a long distance to travel to a health facility should also be targeted by further interventions.
Keywords: Facility delivery, Antenatal care, Determinants, Community health worker, Kenya

* Correspondence: [email protected]
1
JICA SEMAH project, Kisumu, Kenya
2
Graduate School of Biomedical Sciences, Nagasaki University, Nagasaki,
Japan
Full list of author information is available at the end of the article

© 2014 Kawakatsu et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the
Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly credited.
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 2 of 10
http://www.biomedcentral.com/1471-2393/14/265

Background Based on these strategies, community health workers were

More than 500,000 maternal deaths take place annually established to conduct community-based health promotion
worldwide. Most of these deaths occur in low- and activities. A concept known as the Community Health
middle-income countries and could be prevented [1-4]. Strategy was developed and launched in 2006 as health
Research from different countries has shown that the service provision at Level One. The Strategy was well
availability and utilization of skilled birth attendants is a articulated as one of the flagship programs in the National
key factor in reducing maternal mortality [5]. Skilled Health Sector Strategic Plan 2005–2010 (NHPP II), com-
attendance at birth is considered one of the most im- menced in 2006. The Strategy was revised in 2010 and
portant interventions in preventing maternal death. adopted on a national scale. Identified as Level One of
Skilled birth attendants are defined as “accredited health the Kenyan Health System, community health workers
professional[s] such as a midwife, doctor or nurse who (CHWs) in community units (CUs) were required to pro-
has been educated and trained to proficiency in the skills vide primary health services and to create a demand for the
needed to manage normal (uncomplicated) pregnancies, prevention of maternal and childhood deaths. In spite of
childbirth and the immediate postnatal period, and in the government emphasis on facility delivery, there is no re-
identification, management and referral of complications search on the determinants of this type of delivery, includ-
in women and newborns” [4] which implies deliveries ei- ing the CHWs’ working performance, in Kenya.
ther at home or in health facilities. However, the most The aims of the present study are to assess the current
effective strategy for mothers living in lower-income situation of facility delivery, and to investigate factors
countries is to deliver in health facilities under the that influence the utilization of delivery services in
supervision of health professionals [6]. health facilities in rural Kenya using a community-based,
There is a plethora of evidence to show that higher large-population cross-sectional study.
maternal age [7-10], first birth [11], shorter distance to
health facility [12] or availability of transportation [10], Methods
and household wealth or ability to pay [6,7,13] are sig- Research site and population
nificant variables to encourage pregnant women to ac- This study was conducted from August to September,
cess skilled delivery. The use of delivery services can also 2011 in all 64 community units covered by community
be influenced by family composition, religious or trad- health workers (CHWs), who have been identified as
itional beliefs [7,12-16] and maternal occupation [17,18]. Level One of the health system in Kenya since 2006 [24]
Parental education, especially maternal education, is and had been volunteering since May, 2011 in Siaya,
considered one of the strongest factors associated with Ugenya, Gem and Kisumu West districts, Nyanza Province,
receiving trained assistance at delivery [8,11-13]. Ante- Kenya. The MOH identified a six-tier health system, in
natal care visits (ANC) encourage pregnant women to which the Community Unit (level 1) is the proximal imple-
deliver at a health facility [7,19,20]. In addition, the pres- mentation unit to promote primary health care services.
ence of health workers who provide ANC visits at the The range of the population in each CU was from around
community level can also increase the use of skilled 2,000 to 10,000 according to the geographical context. The
attendants [21]. Maternal knowledge about risks of dispensary and health centers (levels 2 and 3 of the Kenyan
delivery and the availability of health services in their health system) are engaged in both preventive and curative
communities is increased by access to information through care. The higher-level hospitals (levels 4, 5 and 6) put more
modern media and maternal/familial education [6,7]. More- focus on curative and rehabilitative aspects than other
over, women with a good knowledge of health issues, such levels [25]. The facilities at level 2 and higher provide health
as the danger signs during pregnancy, are more likely to de- services for childbirth. Twenty-four-hour health services
liver in a health facility [22]. are provided at level 3 and above in the Kenyan health
In Kenya, 42.6% percent of births occurred in a health system.
facility [23]. The proportion of births assisted by medical CHWs were selected and endorsed by the Community
personnel increased slightly from 2003 to 2007 [23]. Ac- Health Committee, which was democratically elected as
cording to Kenya Demographic and Health Survey a governing body of the Community Unit. Their main
(KDHS) [23], traditional birth attendants (TBAs) con- activities are door-to-door canvassing to teach health-
tinue to play an important role in childbirth, where they related preventive methods and collect health-related
assist with 28 percent of births [23]. Twenty-one percent data from each household. This area is mainly inhabited
of births are supported by relatives and friends [23]. The by subsistence farmers and fishermen. The main ethnic
government of Kenya has begun implementing policies to group is Luo and their principal language is the Luo lan-
increase deliveries conducted by skilled health personnel. In guage, followed by Swahili and English. The research
1994, the Kenya Health Policy Framework (KHPF) was population consisted of all mothers who had children
published to pursue the principles of primary health care. aged 12–23 months in this research area.
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 3 of 10
http://www.biomedcentral.com/1471-2393/14/265

The sample size was calculated assuming the following wall, floor and roof were good, we scored one in each
points: that 80% power to demonstrate an odds ratio of variable. This means the maximum score for household
1.4 to be significant at the 5% level, if the ratio exposed: materials was three. The monthly salary was also scored
unexposed is 1:1 and the prevalence of the outcome is in three levels as 0: less than 3,300 KSh; 1: 3,301-5,300
42.6% in the unexposed. This calculated a sample size of KSh; and 2: higher than 5,300 KSh. The sum of the three
1,120. In addition to taking the cluster design effect and variables was calculated and divided into quintiles.
missing data into consideration, the final sample size An indicator on the CHWs’ performance was gener-
was 2,560. ated by using both frequency of visitation to households
and the satisfaction score as reported by the target
Study design and data collection mothers. Frequency of household visitation was scored
A community-based cross-sectional survey was con- as follows:- 0: less than once per month; 1: once per
ducted from August to September, 2011 as a benchmark month; 2: more than once per month. Mothers were also
for the impact assessment of the interventions by the asked to specify their level of satisfaction with the
JICA SEMAH project under authorization from MOH, CHWs’ performance using a five-point Likert scale, di-
Kenya. A total of 11,906 mothers who had children aged vided as low (0), moderate (1) or high (2). Finally, the
12–23 months were identified by CHWs in the 64 variable of the CHWs’ performance was generated by
sub-locations in August, 2011. Forty mothers in each adding the score of the household visitation and the
sub-location were selected using random-sampling score of satisfaction, and categorized into five quintiles
methods. Finally 2,560 mothers were targeted and were as: poorest (0), poor (1), moderate (2), high (3) and high-
asked, using an interviewer-administered questionnaire, to est (4). Social capital was measured using the standard
assess their socio-economic status, their health-seeking questionnaire [26]. In this study, social capital consists
behavior and their sanitation practice. Social capital was of two main areas: informal social control, and social
measured by the standard questionnaire [26]. Most of the cohesion and trust. Each area was represented by a five-
variables were mentioned in the study [6]. In addition, the point Likert scale. The average score of 10 questions was
mothers were asked about the number of times CHWs calculated and classified as: lowest (0), low (1), moderate
visited their household and their satisfaction score regard- (2), high (3) and highest (4).
ing these visits, to generate an indicator of the CHWs’ Media and communication equipment was scored by
performance. possession of radio, TV and cell phone, and classified as:
The research assistants, not CHWs, were recruited from Possession of one piece of equipment or none (0), two
each sub-location. All were high school graduates (12 years pieces of equipment (1) and all three pieces of equipment
of education) and had previous similar experience of data (2). In addition, household sanitation practice was scored
collection. Furthermore, one day of intensive training, by using three indicators, i.e. having a toilet, hand-washing
including guidance in data-collection procedures and facility and a dish rack. They were grouped as: none (0),
pre-testing the questionnaire, was conducted. possession of one out of the three facilities (1), possession
of two (2), and possession of all (3).
Selected study variables
The outcome variable in this study was the place of de- Data storage and analysis
livery for pregnant women. While deliveries at any level Data were verified by a double-entry method and stored
of health facility (dispensary, health center and hospital using Epi Info version 3.5. Statistical analysis was per-
or higher level) were considered institutional deliveries, formed using STATA version 12 (STATA Corporation,
deliveries anywhere other than an institution, including TX, USA). The confidence level was set at 95%. Bivariable
TBA or kinship homes, were considered non-institutional analyses were conducted in order to assess the association
deliveries. between their delivery place and the community, family,
To assess maternal health knowledge, the mothers and individual variables. All 17 independent variables were
were asked about the vaccination schedule, danger signs entered into multiple logistic regression analysis and the
and risk factors in pregnancy and HIV/malaria prevent- final model was selected by a backward elimination strat-
ive methods. After scoring by the Clinical Officer, this egy. The data was weighted with consideration of complex
variable was classified into three levels: low, middle and sampling design during the bivariable and multiple logistic
high. The household wealth index was evaluated using regression analysis.
three variables: household assets (e.g. cell phone, televi- Informed consents from all participants were obtained
sion, bicycle, etc.), house materials for the walls, floor after full explanation of the study design and purposes.
and roof, and monthly salary. If they had more than This research was approved by Great Lake University of
three items out of six household assets, we scored one. Kisumu (GLUK) Ethical Review Committee (GERC) in
If not, we scored zero. If the house materials for the Kenya.
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 4 of 10
http://www.biomedcentral.com/1471-2393/14/265

Results possession of media equipment, number in the house-

Response rate and descriptive statistics hold and number of children under five, household sani-
A total of 2,560 mothers who had children aged 12–23 tation practice, distance from the nearest health facility
months were visited. Thirty-six mothers refused to par- and usual transportation method.
ticipate in this study, 219 mothers were on leave, and We explored the relationship between institutional
the age of 248 children was more than 23 months or less delivery and 17 independent variables using logistic re-
than 12 months. In addition, two children had died be- gression analysis with backward elimination. The results
fore our study started. The data on the place of delivery are also presented in Table 2. Women who finished pri-
of 29 mothers was not available. Thus the final analysis mary (Adjusted Odds ratio (AOR): 1.63 [1.287 – 2.058])
was based on 2,026 mothers. or secondary school and higher (AOR: 2.35 [1.710 –
Table 1 presents the general profile of mothers with 3.241]) are more likely to give birth at a health facility
children aged 12–23 months and is divided into two than women who did not complete primary level or had
groups, institutional and non-institutional delivery. Less no formal education. Higher awareness of health issues
than half of the mothers (48%) had delivery at a health was the factor that was significantly associated with fa-
facility. In the group of mothers who had a facility deliv- cility delivery (AOR: 1.54 [1.113 – 2.129]).
ery, 22.6% of the deliveries occurred in a hospital (level In addition, if women visited ANC services four times
4 or above in the Kenyan health system), 19.3% were in a or more, they were likely to give birth at a health facility
health center (level 3) and 6.32% were at dispensary level (AOR: 1.66 [1.343 – 2.052]). Those who gave birth for the
(level 2). Out of the 51% of mothers who delivered in a first time (AOR: 2.86 [1.588 – 5.158]) had a greater chance
non-institutional location, 10.1% delivered at a TBA’s place, of delivering in a health care setting than women with
13.5% in their home with a TBA, 25.3% in their home with short birth intervals. Compared with women who lived in
a family member and 2% in other non-institutional loca- the poorest households, women who lived in poor (AOR:
tions. Most of the mothers (85%) were married and the edu- 1.42 [1.063 – 1.897]), middle (AOR: 2.10 [1.529 – 2.876]),
cation level for about half of them (58%) was primary rich (AOR: 1.95 [1.286 – 2.950]) and richest (AOR: 2.29
(elementary) level or higher. Fewer than half (46%) of the [1.555 – 3.370]) groups were more likely to have deliveries
mothers had an average health knowledge about danger at a health facility. Similarly, women who possessed two
signs, sanitation practices, etc. More than half of them sanitation facilities (AOR: 2.48 [1.735 – 3.549]) or all the
(55%) had visited health facilities four times or more for facilities (AOR: 2.79 [1.683 – 4.636]) were more likely to
ANC services. About one quarter (26%) of the children aged give birth in health settings, compared to women living in
12–23 months were first-born. As for household character- households with no sanitation facilities. Moreover, women
istics, 21% of them were in the poorest group, 33% were in living with fewer than five household members were more
poor, 22% in middle wealth, 10% of them were in rich and likely to have a facility delivery (AOR: 1.53 [1.190 –
13% were in our highest group. The total number of house- 1.959]). Finally, women who live near a health facility (less
hold members and children under five were mostly more than 20 minutes’ walk) have 2.5 times higher odds of giv-
than five and two respectively. Only 6% of the mothers had ing birth in a medical setting than those who live furthest
no sanitation practice. For 15% it would take more than away (AOR: 2.48 [1.735 – 3.549]).
60 minutes to go to the nearest health facility on foot and
less than 20 minutes for 20% of them. The main means of Discussion
transportation to get to a health facility was on foot. Health facility utilization for delivery services
This study was conducted to identify the determinants of
Factors associated with institutional delivery health facility utilization for deliveries in western Kenya. In
Table 2 shows factors associated with facility delivery. this study, 48% of births occurred as institutional deliveries.
Compared with younger mothers less than 20 years old, Out of those, 23% of the deliveries occurred in hospital
the mothers who were 20–24 and more than 35 years (level 4 or higher), 19.3% were in a health center (level 3)
old had significantly lower odds of having a facility deliv- and 6.3% were in a dispensary (level 2). However, the num-
ery. A higher education level, better awareness of health ber of higher-level health facilities available was less than
issues and an understanding of English were also signifi- the number of low-level facilities, which means that there
cantly associated with facility delivery. If a mother vis- were six hospitals (level 4 or higher), 27 health centers
ited four or more times for ANC, she had more than 1.5 (level 3) and 79 dispensaries (level 2) [27]. This may be due
times the odds of having an institutional delivery. With to fewer health workers in lower-level facilities and poor
first-time pregnancies, the mother had more than twice health services. In addition, level 2 health facilities are
the odds of delivering in health facilities. In addition, closed after 5 pm, but labor frequently begins at night.
other factors that had a significant effect on facility On the other hand, 52% of deliveries occurred in non-
delivery at the bivariate level were household wealth, institutional locations, such as a TBA’s place (10.1%), the
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 5 of 10
http://www.biomedcentral.com/1471-2393/14/265

Table 1 Socio-demographic characteristics of mothers Table 1 Socio-demographic characteristics of mothers

with children aged 12–23 months and their community with children aged 12–23 months and their community
variables by place of delivery variables by place of delivery (Continued)
Variables Institutional Non-institutional Total Birth interval
N (%) N (%) N (%)
Short birth interval 30 (3.1) 51 (4.9) 81
Place of delivery 976 (48.2) 1050 (51.8) 2026 (<24 months) (4.0)
(100)
Medium birth interval 78 (8.0) 107 (10.2) 185
Age group (years) (24–47 months) (9.1)
< 20 90 (9.2) 76 (7.2) 166 Long birth interval 539 (55.2) 690 (65.7) 1229
(8.2) (>48 months) (60.7)
20 – 24 301 (30.8) 340 (32.4) 641 First birth 329 (33.7) 202 (19.2) 531
(31.6) (26.2)
25 – 29 244 (25.0) 246 (23.4) 490 Household wealth index
(24.2)
Poorest 154 (16.3) 272 (26.6) 426
30 – 34 160 (16.4) 159 (15.1) 319 (21.7)
(15.8)
Poor 299 (31.6) 350 (34.3) 649
≥ 35 181 (18.6) 229(21.8) 410 (33.0)
(20.2)
Middle 226 (23.9) 213 (20.9) 439
Marital status (22.3)
Single/Divorced/Widowed 153 (15.7) 150 (14.3) 303 Rich 110 (11.6) 88 (8.6) 198
(15.0) (10.1)
Married 823 (84.3) 899 (85.7) 1722 Richest 157 (16.6) 98 (9.6) 255
(85.0) (13.0)
Education level Media and communication
No education/dropped 312 (32.0) 530 (50.5) 842 equipment
out from primary (41.6) Possession of one item or 277 (28.4) 416 (39.7) 693
Primary level 421 (43.1) 396 (37.7) 817 none (34.2)
(40.3) 2 items 544 (55.7) 545 (52.0) 1089
Secondary level or higher 243 (24.9) 124 (11.8) 367 (53.8)
(18.1) All items 155 (15.9) 88 (8.4) 243
Occupation (12.0)
Subsistence farmer 528 (55.5) 597 (57.3) 1125 Number of household
(56.4) members
Domestic worker/student 94 (9.9) 114 (10.9) 208 5 or more 627 (64.2) 772 (73.5) 1399
(10.4) (69.1)
Other skilled employee 330 (34.7) 331 (31.8) 661 Less than 5 349 (35.8) 278 (26.5) 627
(33.2) (31.0)
Literacy in English Number of children under
five
Illiterate 169 (17.4) 309 (29.4) 478
(23.6) 2 or more 584 (59.8) 736 (70.1) 1320
(65.2)
Literate 804 (82.6) 741 (70.6) 1545
(76.4) Less than 2 392 (40.2) 314 (29.9) 706
(34.9)
Maternal health
knowledge Household sanitation
practice
Low 116 (11.9) 199 (19.0) 315
(15.6) No sanitation facilities 33 (3.4) 95 (9.1) 128
(6.3)
Middle 426 (43.7) 506 (48.3) 932
(46.1) Possession of at least one 320 (32.8) 464 (44.2) 784
(38.7)
High 432 (44.4) 343 (32.7) 775
(38.3) Possession of two 319 (32.7) 281 (26.7) 600
(29.6)
ANC visits
Possession of all 304 (31.2) 210 (20.0) 514
Less than 4 times 358 (36.7) 544 (51.8) 902 (25.4)
(44.5)
4 times or more 618 (63.3) 506 (48.2) 1124
(55.5)
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 6 of 10
http://www.biomedcentral.com/1471-2393/14/265

Table 1 Socio-demographic characteristics of mothers visit, birth interval, socio-economic status of the household,
with children aged 12–23 months and their community family size, household sanitation practices and proximity of
variables by place of delivery (Continued) the nearest health facility.
Traveling time to nearest Maternal education as one of the significant determi-
health facility nants of institutional delivery has been well-established
More than 60 min 125 (12.8) 185 (17.6) 310 and evidenced within several studies [6,28,29]. Higher
(15.3) levels of education improve maternal knowledge, in-
41 – 60 min 257 (26.3) 304 (29.0) 561 crease self-confidence and also enhance awareness of
(27.7) available health resources in the community. Female
21 – 40 min 349 (35.8) 401 (38.2) 750 decision-making abilities and problem-solving skills
(37.0)
are also strengthened by receiving a higher education
20 min or less 245 (25.1) 160 (15.2) 405 and change the household dynamics [8,28,30,31].
(20.0)
Better maternal health knowledge was also a signifi-
Usual transportation cant determinant of facility delivery in this study. A
On foot 822 (84.4) 949 (90.5) 1771 study found that women in Zambia who can pinpoint
(87.5)
the danger signs in pregnancy are more likely to deliver
Any other means of 152 (15.6) 100 (9.5) 252 in a health facility than those without such knowledge
transportation (12.5)
[22]. High maternal health knowledge may be able to
Performance of CHWs positively influence a woman’s care-seeking behaviors as
Poorest 357 (36.6) 373 (35.6) 730 well as enabling her to recognize the danger signs early.
(36.1)
However, this maternal knowledge may be the result of
Poor 82 (8.4) 103 (9.8) 185 frequent contacts with skilled health personnel; therefore
(9.2)
it is necessary to conduct a prospective study.
Middle 200 (20.5) 193 (18.4) 393 In addition, receiving antenatal care four times or
(19.4)
more was one of the significant positive determinants to
High 182 (18.7) 230 (22.0) 412
(20.4)
improving institutional delivery in this study. Most of
the studies have found that women who use ANC ser-
Highest 154 (15.8) 148 (14.1) 302
(14.9) vices are much more likely to receive skilled attendants
at delivery [19,21,32]. Women who visit health facilities
Social capital
for ANC services receive more opportunities for health
Lowest 171 (17.5) 194 (18.5) 365
(18.0)
professionals to explain the advantages of skilled attendants
and information on the status of their pregnancies. Other
Low 193 (19.8) 185 (17.6) 378
(18.7) indicators or markers of a maternal health-seeking behavior
may include ANC attendance or being able to interact with
Middle 229 (23.5) 263 (25.1) 492
(24.3) the health system and health facility in a more comfortable
High 217 (22.2) 211 (20.1) 428
manner [6].
(21.1) Women in their first pregnancy are more likely to
Highest 166 (17.0) 197 (18.8) 363 have their baby in a health facility [33]. The first birth is
(18.0) known to be more difficult and the family may help the
mother to get better care [6]. Contrasting with this is
the fact that women of higher parity tend to rely on their
mother’s home with a TBA assistant (13.5%), the mother’s previous experiences and may not think of the necessity
home with family and friends (25.3%) and other places for health care services if previous deliveries were un-
(2%). Despite the fact that TBAs are not officially complicated [34].
recognized as qualified caregivers, nearly half of the Household wealth was also found to be an important
non-institutional deliveries were assisted by TBAs and determinant of facility delivery. We noted that there was
family members. This result implies that community a strong correlation between the wealth index and the
members still demand the informal health service use of maternal care services. This is consistent with
provision available in their residential areas and rely on findings from the previous studies [28]. Households on
conventional kinship and social network. quite a limited budget could have difficulty paying fees
and therefore would tend to be less likely to use a health
Determinants of institutional delivery facility for delivery. In addition, women of low economic
The significant determinants of institutional delivery were: status are known to have lower rates of seeking maternal
maternal education level, maternal health knowledge, ANC health care. A greater use of services is more commonly
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 7 of 10
http://www.biomedcentral.com/1471-2393/14/265

Table 2 Factors associated with institutional delivery at Table 2 Factors associated with institutional delivery at
64 CUs in Kisumu West, Siaya, Ugenya and Gem districts, 64 CUs in Kisumu West, Siaya, Ugenya and Gem districts,
Nyanza Province, Kenya Nyanza Province, Kenya (Continued)
Variables Bivariable analysis Multivariate Birth interval
analysis
Short birth interval Ref. Ref.
Unadjusted 95% Adjusted 95% (<24 months)
odds ratio (OR) CI OR CI
Medium birth interval 1.234 0.687- 1.475 0.784-
Age group (years)
(24–47 months) 2.218 2.776
< 20 Ref.
Long birth interval 1.288 0.778- 1.646 0.953-
20 – 24 *0.628 0.430- (>48 months) 2.131 2.844
0.917
First birth **2.426 1.434- ***2.862 1.588-
25 – 29 0.740 0.502- 4.104 5.158
1.091
Household wealth
30 – 34 0.735 0.486- index
1.111
Poorest Ref. Ref.
≥ 35 **0.564 0.380-
Poor *1.505 1.143- *1.420 1.063-
0.837
1.981 1.897
Marital status
Middle ***2.164 1.608- ***2.097 1.529-
Single/Divorced/ Ref. Ref. 2.911 2.876
Widowed
Rich ***2.392 1.629- **1.948 1.286-
Married 0.908 0.695- 1.219 0.892- 3.512 2.950
1.187 1.665
Richest ***3.391 2.382- ***2.290 1.555-
Education level 4.825 3.370
No education/dropped Ref Ref. Media and
out from primary communication
equipment
Primary level ***1.886 1.522- ***1.628 1.287-
2.338 2.058 Possession of one item Ref.
or none
Secondary level or ***3.427 2.567- ***2.354 1.710-
higher 4.540 3.241 2 items ***1.494 1.207-
1.849
Occupation
All items ***2.741 1.974-
Subsistence farmer Ref. 3.807
Domestic worker/ 1.019 0.746- Number of household
student 1.394 members
Other skilled employee 1.232 0.995- 5 or more Ref. Ref.
1.525
Less than 5 ***1.534 1.245- **1.527 1.190-
Literacy in English 1.890 1.959
Illiterate Ref. Number of children
Literate ***1.984 1.573- under five
2.502 2 or more Ref. Ref.
Maternal health Less than 2 ***1.482 1.210- 1.137 0.887-
knowledge 1.815 1.457
Low Ref. Ref. Household sanitation
Middle 1.271 0.956- 1.130 0.829- practice
1.691 1.540 No sanitation facilities Ref. Ref.
High ***1.955 1.458- **1.539 1.113- Possession of at least **1.879 1.191- 1.602 0.984-
2.622 2.129 one 2.964 2.606
ANC visits Possession of two ***3.015 1.895- ***2.480 1.511-
Less than 4 times Ref. Ref. 4.798 4.071
4 times or more ***1.848 1.523- ***1.660 1.343- Possession of all ***3.904 2.438- ***2.793 1.683-
2.243 2.052 6.252 4.636
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 8 of 10
http://www.biomedcentral.com/1471-2393/14/265

Table 2 Factors associated with institutional delivery at would be associated with health awareness/practices in the
64 CUs in Kisumu West, Siaya, Ugenya and Gem districts, household and the level of available expenditure for health
Nyanza Province, Kenya (Continued) amenities.
Traveling time to Finally, a number of studies have examined the effect
nearest health facility of distance, quality and transportation means [14,21,32].
More than 60 min Ref. Ref. It is clear that transportation issues are a major factor in
41 – 60 min *1.394 1.021- 1.357 0.968- the decision to go to a health facility (phase 1 delay).
1.904 1.903 This factor also influences the delay caused by distance
21 – 40 min *1.373 1.023- 1.324 0.961- from home to the health facility (phase 2 delay). In our
1.844 1.826 study, the mode of travel to the nearest facility was not
20 min or less ***2.560 1.845- ***2.482 1.735- significantly associated with facility delivery. Most labor
3.554 3.549 began in the late evening, and at that time it was difficult
Usual transportation to find the usual transportation in the research area.
On foot Ref. Ref. In our study, performance of CHWs and social capital
Any other means of ***1.836 1.368- 1.379 0.992- as community variables were not significant determi-
transportation 2.466 1.918 nants of institutional delivery. Even though one of the
Performance of CHWs CHWs’ main activities was household visitation to pro-
Poorest Ref.
vide health education to community members, there was
no effect on facility delivery. A study in Nepal also found
Poor 0.776 0.543-
1.109 that household visitation by health workers was associ-
ated with the increased utilization of ANCs and post-
Middle 1.158 0.884-
1.516 natal care. However, this factor did not have a significant
High 0.796 0.610-
impact on skilled birth attendance [36]. Recent evidence
1.039 from Tanzania suggests that an increase in home visits
Highest 1.033 0.770- by health personnel at the grass roots level can increase
1.387 the use of ANC visits as well as delivery by a skilled birth
Social capital attendant [37]. The intervention of the study in Tanzania
Lowest Ref.
was to train safe motherhood promoters (SMPs) equivalent
to CHWs in Kenya. Their roles are focused on promoting
Low 1.119 0.816-
1.535 early ANCs, complete ANCs and delivery with a skilled at-
tendant. Compared with SMPs in Tanzania, CHWs in
Middle 1.005 0.754-
1.341 Kenya have multiple tasks focusing not only on antenatal
High 1.117 0.825-
care, facility delivery and postnatal care but also case man-
1.510 agement for the general population, heavy reporting, data
Highest 0.855 0.634- management, community entrepreneurship, and so on. It
1.153 would be better to give CHWs limited, clear objectives and
*< 0.05. knowledge to promote community health and reduce the
**< 0.01. amount of reporting duty. This would make the work of
***< 0.001.
CHWs more effective, enhance community awareness and
increase utilization of ANC visits and delivery by skilled
seen in households with higher economic status where health personnel. In addition, it would be necessary to
modern health care services are easily accessible [11]. develop hands-on tools about the merits of institutional
There is a significant association with family size and delivery and the risks of non-institutional delivery for
facility delivery in this study. Stephen et al. [35] argue the CHWs to work more effectively.
that families in the area with high birth rates may be
more conservative toward antenatal and maternal health Strengths and limitations
services. A large family size might be a reflection of the This study was a large community-based cross-sectional
degree of a woman’s independence. Additionally, the study with adequate power to detect important differ-
decision-maker in the household with a large family size ences in facility delivery between different groups. Rec-
is usually the father. Further research should include the ommended by Sabine Gabrysch et al. [6], seventeen
indicator on these factors, so that the effect of family comprehensive individual, households and community
size will become a more concrete indication. factors were assessed to adjust confounding factors.
In households that have sanitation facilities, women are This study has some limitations that should be noted.
more likely to give birth at a health facility. This indicator Firstly, the participants provided all of the data. There
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 9 of 10
http://www.biomedcentral.com/1471-2393/14/265

was no way to validate the obtained information with Author details

1
any objective source such as a health facility card. Al- JICA SEMAH project, Kisumu, Kenya. 2Graduate School of Biomedical
Sciences, Nagasaki University, Nagasaki, Japan. 3Ministry of Public Health and
though self-reporting can be sketchy, it can be assumed Sanitation, Siaya, Kenya. 4Ministry of Public Health and Sanitation, Kisumu
that biases are less likely in pregnancy-related events as West, Kenya.
compared to more sensitive issues, such as sexual behav-
Received: 30 August 2012 Accepted: 5 August 2014
ior and drug abuse. Secondly, we examined the associ- Published: 9 August 2014
ation between current situations and previous places of
delivery. This limitation is especially related to CHWs’
References
status and social capital. To minimize these limitations,
1. Lawn JE, Cousens S, Zupan J: 4 million neonatal deaths: When? Where?
it would be better for further research to select pregnant Why? Lancet 2005, 365(9462):891–900.
women before their delivery and the use of a prospective 2. Ronsmans C, Graham WJ: Maternal mortality: who, when, where, and
study would be needed to examine causal relationship. why. Lancet 2006, 368(9542):1189–1200.
3. Stanton C, Lawn JE, Rahman H, Wilczynska-Ketende K, Hill K: Stillbirth rates:
Although we examined seventeen comprehensive inde- delivering estimates in 190 countries. Lancet 2006, 367(9521):1487–1494.
pendent variables, we missed some variables in this re- 4. AbouZahr C, Wardlaw T: Maternal Mortality in 2000: Estimates Developed by
search, such as parity and time of onset of labor. In WHO, UNICEF and UNFPA. Geneva: WHO; 2003.
5. Koblinsky MA, Campbell O, Heichelheim J: Organizing delivery care: what
addition, it was difficult to get accurate data on the preg- works for safe motherhood? Bull World Health Organ 1999, 77(5):399–406.
nancy or delivery complications in this research. There- 6. Gabrysch S, Campbell OMR: Still too far to walk: literature review of the
fore, it would be better for future research to include determinants of delivery service use. BMC Pregnancy Childbirth 2009, 9:34.
7. Rai RK, Singh PK, Singh L: Utilization of maternal health care services
these variables and to consider clear definitions of the among married adolescent women: insights from the Nigeria demographic
variables. and health survey, 2008. Women’s Health Issues 2012, 22(4):e407–e414.
8. Burgard S: Race and pregnancy-related care in Brazil and South Africa.
Soc Sci Med 2004, 59(6):1127–1146.
Conclusion 9. Reynolds HW, Wong EL, Tucker H: Adolescents’ use of maternal and
child health services in developing countries. Int Fam Plan Perspect 2006,
Despite the limitations, the determinants we found have 32(1):6–16.
useful implications for both health care providers and 10. Edmonds JK, Paul M, Sibley L: Determinants of place of birth decisions in
decision-makers in health programs. The results suggest uncomplicated childbirth in Bangladesh: An empirical study. Midwifery
2012, 28(5):554–560.
that the involvement of TBAs to promote facility deliv- 11. Navaneetham K, Dharmalingam A: Utilization of maternal health care
ery is still one of the most effective strategies. Strength- services in Southern India. Soc Sci Med 2002, 55(10):1849–1869.
ening CHWs’ performance with clear guidance and 12. Thaddeus S, Maine D: Too far to walk: maternal mortality in context.
Soc Sci Med 1994, 38(8):1091–1110.
hands-on tools might also be an effective intervention. 13. Kamiya Y, Yoshimura Y, Islam MT: An impact evaluation of the safe
Stressing the importance of regular attendance at ANC motherhood promotion project in Bangladesh: evidence from Japanese
(more than four times) would be effective to enhance aid-funded technical cooperation. Soc Sci Med 2013, 83:34–41.
14. Glei DA, Goldman N, Rodríguez G: Utilization of care during pregnancy
motivation for a facility delivery. Based on our findings, in rural Guatemala: does obstetrical need matter? Soc Sci Med 2003,
those actions to improve the facility delivery rate should 57(12):2447–2463.
be focused more on those pregnant women who have a 15. Kyomuhendo GB: Low use of rural maternity services in Uganda: impact
of women’s status, traditional beliefs and limited resources. Reprod Health
low education level and immature health knowledge. In Matters 2003, 11(21):16–26.
addition, women who have short birth intervals and live 16. Mrisho M, Schellenberg JA, Mushi AK, Obrist B, Mshinda H, Tanner M,
in households with poor economic status, a large num- Schellenberg D: Factors affecting home delivery in rural Tanzania. Trop
Med Int Health 2007, 12(7):862–872.
ber of family members and are far from the nearest 17. Nwakoby BN: Use of obstetric services in rural Nigeria. J R Soc Health
health facility should also be targeted. 1994, 114(3):132–136.
18. Addai I: Determinants of use of maternal-child health services in rural
Ghana. J Biosoc Sci 2000, 32(1):1–15.
Competing interests 19. Paul BK, Rumsey DJ: Utilization of health facilities and trained birth
The authors declare that they have no competing interests. attendants for childbirth in rural Bangladesh: an empirical study. Soc Sci
Med 2002, 54(12):1755–1765.
Authors’ contributions 20. Adjiwanou V, LeGrand T: Does antenatal care matter in the use of skilled
YK was the responsible person for conducting the study and writing the birth attendance in rural Africa: A multi-country analysis. Soc Sci Med
manuscript. TS was an overall supervisor of this research and gave technical 2013, 86:26–34.
advice. KO, SW, KK and EW managed the data collection process in four 21. Gage AJ, Guirlène Calixte M: Effects of the physical accessibility of
districts. SH advised on statistical analysis and gave approval to submit. All maternal health services on their use in rural Haiti. Popul Stud 2006,
authors read and approved the final manuscript. 60(3):271–288.
22. Stekelenburg J, Kyanamina S, Mukelabai M, Wolffers I, Van Roosmalen J:
Waiting too long: Low use of maternal health services in Kalabo,
Acknowledgements Zambia. Trop Med Int Health 2004, 9(3):390–398.
The JICA SEMAH Project supported and funded this survey as part of their 23. Kenya National Bureau of Statistics (KNBS): 2008–09 Kenya demographic and
project activities. DHMT members in our research area supported this study. health survey. Nairobi, Kenya: 2010.
The authors would like to thank all of the people who supported this 24. James L, Ole-Kiyiapi MKB: Strategic plan for health information system
research, with special thanks to the Ministry of Health, Kenya and all 2009–2014. In Ministry of medical services, ministry of public health and
participants. sanitation. Nairobi, Kenya: 2009.
Kawakatsu et al. BMC Pregnancy and Childbirth 2014, 14:265 Page 10 of 10
http://www.biomedcentral.com/1471-2393/14/265

25. Ministry of Health , Kenya: Reversing the Trends: The Second National Health
Sector Strategic Plan of Kenya- NHSSP II(2005-2010). Nairobi, Kenya: 2005.
26. Sampson RJ, Raudenbush SW, Earls F: Neighborhoods and violent crime:
a multilevel study of collective efficacy. Science 1997, 277(5328):918–924.
27. EHealth-Kenya facilities. [http://www.ehealth.or.ke/facilities/default.aspx]
28. Sagna ML, Sunil TS: Effects of individual and neighborhood factors on
maternal care in Cambodia. Health Place 2012, 18(2):415–423.
29. Stella B, Adesegun F: Determinants of use of maternal health services in
Nigeria - looking beyond individual and household factors. BMC
Pregnancy Childbirth 2009, 9(43) doi:10.1186/1471-2393-9-43.
30. Furuta M, Salway S: Women’s position within the household as a
determinant of maternal health care use in Nepal. Int Fam Plan Perspect
2006, 32(1):17–27.
31. Raghupathy S: Education and the use of maternal health care in
Thailand. Soc Sci Med 1996, 43(4):459–471.
32. Gage AJ: Barriers to the utilization of maternal health care in rural Mali.
Soc Sci Med 2007, 65(8):1666–1682.
33. Agha S, Carton TW: Determinants of institutional delivery in rural Jhang,
Pakistan. Int J Equity Health 2011, 10. doi:10.1186/1475-9276-10-31.
34. Stephenson R, Tsui AO: Contextual influences on reproductive health
service use in Uttar Pradesh, India. Stud Fam Plann 2002, 33(4):309–320.
35. Stephenson R, Baschieri A, Clements S, Hennink M, Madise N: Contextual
influences on the use of health facilities for childbirth in Africa. Am J
Public Health 2006, 96(1):84–93.
36. Sharma SK, Sawangdee Y, Sirirassamee B: Access to health: Women’s status
and utilization of maternal health services in Nepal. J Biosoc Sci 2007,
39(5):671–692.
37. Mushi D, Mpembeni R, Jahn A: Effectiveness of community based safe
motherhood promoters in improving the utilization of obstetric care.
The case of Mtwara Rural District in Tanzania. BMC Pregnancy Childbirth
2010, 10. doi:10.1186/1471-2393-10-14.

doi:10.1186/1471-2393-14-265
Cite this article as: Kawakatsu et al.: Determinants of health facility
utilization for childbirth in rural western Kenya: cross-sectional study.
BMC Pregnancy and Childbirth 2014 14:265.

Submit your next manuscript to BioMed Central

and take full advantage of:

• Convenient online submission

• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution

Submit your manuscript at

www.biomedcentral.com/submit