Individual and Community Level Factors Associated With Health Facility Delivery: A Cross Sectional Multilevel Analysis in Bangladesh
Individual and Community Level Factors Associated With Health Facility Delivery: A Cross Sectional Multilevel Analysis in Bangladesh
Individual and Community Level Factors Associated With Health Facility Delivery: A Cross Sectional Multilevel Analysis in Bangladesh
1 The University of Sydney, Faculty of Medicine and Health, School of Public Health, NSW, Australia,
2 International Centre for Diarrhoeal Disease Research, Bangladesh, Dhaka, Bangladesh, 3 Health Nutrition
and Population Programme, BRAC, Dhaka, Bangladesh
* [email protected]
a1111111111
a1111111111
a1111111111 Abstract
a1111111111
a1111111111
Introduction
Improving maternal health remains one of the targets of sustainable development goals. A
maternal death can occur at any time during pregnancy, but delivery is by far the most dan-
OPEN ACCESS
gerous time for both the woman and her baby. Delivery at a health facility can avoid most
Citation: Huda TM, Chowdhury M, El Arifeen S, maternal deaths occurring from preventable obstetric complications. The influence of both
Dibley MJ (2019) Individual and community level
factors associated with health facility delivery: A
individual and community factors is critical to the use of health facility delivery services. In
cross sectional multilevel analysis in Bangladesh. this study, we aim to examine the role of individual and community factors associated with
PLoS ONE 14(2): e0211113. https://doi.org/ health facility-based delivery in Bangladesh.
10.1371/journal.pone.0211113
Conclusion
Our results strongly suggest factors at both Individual, and community level influenced the
use of health facility delivery services in Bangladesh. Thus, any future strategy to improve
maternal health in Bangladesh must consider community contexts and undertake multi-sec-
torial approach to address barriers at different levels. At the individual level the programs
should also focus on the need of the young mother, the multiparous the less educated and
women in the poorest households.
Introduction
Global maternal mortality ratio has fallen by nearly 44% between 1990 and 2015[1]. Despite the
significant reduction over the last two decades, the mortality ratio is still unacceptably high in
many low and middle-income countries. In 2015, approximately 5500 women died (95% CI
3900 to 8800) in Bangladesh from maternal causes[2]. The lifetime risk of women to die from
maternal causes is estimated to be 1 in 240 in Bangladesh[2]. Improving maternal health thus
remains one of the targets of Sustainable Development Goals (SDG). SDG has set a goal of
reducing the maternal mortality ratio (MMR) to less than 70 per 100,000 live births by 2030[3].
As the delivery process can result in unexpected complications, health facility delivery or
delivery by a skilled attendant is crucial. It is now well established that ensuring skilled attendant
at birth, or health facility delivery, can avoid most maternal deaths occurring from preventable
obstetric complications and thus can make a critical difference to the survival of the mother[4,
5]. An analysis of secondary data from 48 low and middle-income countries reported that in
Sub Saharan Africa, South Asia, and Southeast Asia, more than 70% of all births in the lowest
two wealth quintiles occurred at home [6]. In Bangladesh only around 37% of women delivered
in a designated health care facility[7]. Critical to improving these rates is an understanding of
the multilevel factors associated with utilisation of health facility delivery services.
Several studies conducted in Bangladesh and other similar settings have attempted to iden-
tify the determinants of health facility delivery. Most studies have focused on individual and
health system factors and demonstrated a significant effect of those factors on the use of health
facility delivery services[8–13]. Although studies have reported similar sets of determinants,
the effect size differs from one geographic area to another. Therefore, it is possible that unob-
served community factors also influenced the location where women deliver. Few studies have
already documented the role of community or social factors in the utilisation of maternal
health services in South Asia and Africa[14, 15]. However, the role of community factors on
the utilisation of health facility delivery services is still less understood in Bangladesh. We,
therefore, planned to examine a range of individual and community factors and measure their
extent of influence on health facility-based delivery in Bangladesh.
Methods
Data sources
We used data from 2010 Bangladesh Maternal Mortality Survey[16], the largest national
household survey designed to provide the national estimate of maternal mortality ratio
(MMR) and information on family planning, antenatal, delivery, postnatal, and emergency
obstetric care.
The sampling frame for the BMMS survey was divided into urban and rural areas. The pri-
mary sampling unit (PSU) for the urban and rural areas was the ward and union respectively. For
each selected PSU, two mohallas (the next administrative unit for urban area) or a mouza (the
next administrative unit for rural area) were randomly selected and segmented into clusters. A
cluster was then randomly selected from each selected mohallas or mouza. A total of 2,708 clus-
ters were selected including 1,142 urban and 1,566 rural clusters. From these clusters, 175,600
households (around 65 household from each cluster) were then randomly selected for the survey,
of which 168,629 were successfully interviewed. From these 168,629 households, 175,621 women
were interviewed for the measurement of maternal mortality. Among them, information on ante-
natal, delivery and postnatal care were collected from 28,032 women who had a birth in the five
years preceding the survey. We have included all 28,032 women in our study.
Variables
Outcome variable. Our outcome variable is whether a mother delivered at home or in a
health facility. A birth is categorised as health facility-based if it occurred at a private, public or
non-governmental clinic.
Explanatory variables. We selected explanatory variables based on Andersen’s health-
seeking behavioral model. According to Andersen’s behavioral model, predisposing, enabling,
and need factors at the individual and community levels are responsible for increasing health-
seeking behavior and health facilities utilization[17–19]. We have included need factors at the
individual level while predisposing and enabling factors were included at community level.
For community-level factors, we have developed several binary variables by aggregating the
individual level characteristics at the cluster level.
Individual level factors. Age at birth (less than 20, 20–34 years, 35 years and above)
1. Maternal Education (No education, primary incomplete, primary completed, higher). Pri-
mary complete is defined as completing grade 5 and secondary complete is defined as com-
pleting grade 10.
2. Religion (Islam, Hinduism and others)
3. Parity (1, 2, 3 and 4 or more),
4. Maternal care seeking practices (at least 3 ANC or any ANC from a medically trained pro-
vider that is a qualified doctor, nurse, midwife, paramedic, community skilled birth atten-
dant (CSBA) and others as designated by Govt of Bangladesh[7]),
5. Exposure to mass media (watch TV or read the newspaper at least once in a week)
6. Pregnancy complications in her last pregnancy (convulsion/fits, High BP or Edema, Severe
bleeding, mal-presentation and prolonged labour).
7. Household wealth index as a proxy for the women socioeconomic condition.
3. High concentration of exposure to mass media in the community (whether or not more
than 50% population of the cluster read the newspapers at least once in a week or watch TV
at least once in a week),
4. High concentration of use of ANC in the community (whether or not more than 50% pop
of the cluster had at least 1 ANC check from a medically trained provider or had at least 3
ANC checks),
5. High concentration of educated women in the community (whether or not more than 50%
pop of the cluster had at least eight years of education)
6. High concentration of wealth in the community (whether or not more than 50% of the pop-
ulation are in the top 3 wealth quintiles)
Statistical analysis
We performed multilevel regression analyses to assess the individual, household and commu-
nity level factors associated with health facility delivery. We used random effects logistic model
(also known as the mixed effect or random intercept model) with two levels to assess the influ-
ence of individual and community factors on the use of health facility-based delivery services.
Multilevel modelling technique was used to take account the hierarchical structure of our data.
In our survey, women were nested within households and households were nested within clus-
ters. We have considered clusters as our random effect to account for the unexplained variabil-
ity at the community level. All analysis was done on weighted data.
We first constructed an empty” model (model i), which only includes a random intercept.
An empty random effect model will provide an estimation of the degree of correlation in the
health facility delivery that exists at the community level (cluster). We then included all indi-
vidual factors in the model (model ii). Finally, we added the community level factors (model
iii) to examine which contextual factors have the most influence on the use of a health facility
for delivery care. For all models, we presented the odds ratio and associated 95% confidence
intervals. We did all statistical analyses using the Stata statistical software, version 15.
Ethics
The present study relied upon secondary analysis of anonymous, publicly available household
survey data from BMMS 2010. The BMMS 2010 survey was approved by the Ethical Review
Committee (ERC) of the Bangladesh Medical Research Council (BMRC). All study partici-
pants gave informed consent before participation. The raw data of BMMS 2010 is publicly
available. We have downloaded the data with permission from the Measure Evaluation.
Results
We included 28,032 mothers in our analysis. We presented the percentage of women using
health facility delivery by the individual, characteristics in Table 1. The overall use of health
facility delivery services in our sample was 21.6% (95% CI 20.8%, 22.5%). The results showed
significant differences in the use of health facility-based delivery services between catergoreis
of maternal age, education, religion, parity, exposure to mass media, household wealth and
complications experienced during pregnancy. The utilisation of health facility delivery service
was higher among the younger, educated and women from affluent households. The rate was
slightly lower among the Muslim women. Also, women who reported three or more ANC visit
from any provider had a higher rate of health facility delivery than women who only reported
one ANC visit. Similarly, women who reported ANC from a medically trained provider had a
higher rate of health facility delivery than women who reported ANC from a non-medically
trained provider. Also, women with exposure to mass media had a higher rate of health facility
delivery.
Table 2 showed the percentage of women using health facility delivery by community level
factors. Urban women were more likely to deliver in a health facility compared to rural
women. Women residing in communities with a higher concentration of educated mothers,
affluent households, women who have access to media and women who reported to use ANC
were also more likely to deliver in a health facility.
https://doi.org/10.1371/journal.pone.0211113.t003
The intra-class (ICC) correlation in the empty model for health facility delivery is 0.286
(95% CI 0.266, 0.306). The ICC indicates a considerable between cluster heterogeneity. A little
less than one-third of the total variance in health facility delivery was attributable to the differ-
ences across the cluster or community-level factors (ICC). The variations across clusters
remained statistically significant, even after controlling for all factors in the full model.
Discussion
We found along with individual factors, community factors also have a significant influence
on the use of health facility delivery services in Bangladesh. It thus confirms the findings of a
Table 4. Multilevel logistic regression analysis of individual, household and community level factors associated
with health facility delivery.
Model II Model III
AOR (95% CI) AOR (95% CI)
Maternal Age
<20 1 1
20–34 1.27 (1.14, 1.42) 1.26 (1.13, 1.41)
35+ 1.73 (1.47, 2.03) 1.64 (1.38, 1.95)
Parity
1 1 1
2 2.67 (2.35, 3.03) 2.55 (2.23, 2.92)
3 1.69 (1.50, 1.90) 1.61 (1.42, 1.82)
>=4 1.34 (1.19, 1.52) 1.30 (1.14, 1.48)
Religion
Hindu 1 1
Muslim 0.58 (0.40, 0.83) 0.43 (0.30, 0.63)
Others� 0.89 (0.61, 1.30) 0.66 (0.45, 0.98)
Maternal Education
No Education 1 1
Incomplete Primary 1.01 (0.89, 1.14) 1.01 (0.89, 1.15)
Completed Primary 1.01 (0.89, 1.15) 1.03 (0.90, 1.18)
Secondary or Higher 1.42 (1.26, 1.59) 1.43 (1.27, 1.61)
ANC from a Medically Trained Provider
No 1 1
Yes 2.77 (2.56, 3.00) 2.59 (2.38, 2.83)
At least 3 ANC from any provider
No 1 1
Yes 2.05 (1.91, 2.20) 1.91 (1.77, 2.06)
Watch TV daily
No 1 1
Yes 1.35 (1.24, 1.45) 1.13 (1.04, 1.23)
Read Newspaper at least weekly
No 1 1
Yes 1.40 (1.28, 1.53) 1.40 (1.27, 1.54)
Complication
Headache & blur vision 1.22 1.38 (1.26, 1.52)
(1.12, 1.33)
Convulsion/fits 2.01 (1.52, 2.67) 2.20 (1.63, 2.98)
High Blood Pressure 3.15 (2.57, 3.85) 3.35 (2.70, 4.16)
Severe heavy bleeding 2.58 (1.91, 3.49) 2.80 (2.05, 3.83)
Leaking membrane 3.70 (3.26, 4.21) 3.89 (3.41, 4.44)
Oedema 1.19 (1.08, 1.31) 1.24 (1.12, 1.38)
None of the above 1 1
Wealth Index
Poorest 1 1
Second 1.10 (0.97, 1.25) 1.13 (0.99, 1.30)
Middle 1.45 (1.28, 1.64) 1.53 (1.34, 1.74)
Fourth 1.70 (1.51, 1.93) 1.79 (1.57, 2.04)
Richest 2.75 (2.41, 3.15) 3.15 (2.72, 3.65)
(Continued )
Table 4. (Continued)
https://doi.org/10.1371/journal.pone.0211113.t004
previous study and reiterate the importance of community factors with respect to use of health
facility delivery services[20].
Similar to other studies we found a high concentration of wealthier households in a com-
munity positively influences health facility delivery in that community[14, 21]. It is possible in
communities where there is a high concentration of wealthier households, health facility deliv-
ery practice may become a norm, that other women from poorer households may follow[21].
It is also possible that in communities where there is a high concentration of wealth, health
facilities function better and provide quality services, which in turn, can have a positive influ-
ence on the overall health service utilisation in the community. Our findings suggest women
socioeconomic condition has a strong positive influence on health facility delivery[11, 13, 22].
These findings indicated delivering at a health health facility is influenced by the economic
resources available to an individual.
Education both at individual and community level exerts a positive influence on the overall
health service utilisation within the community. Several studies in Bangladesh and elsewhere
have reported a strong association between women’s education and use of health facility deliv-
ery and other maternal health services [23–27]. Formal education can influence the use of
health facility delivery in multiple pathways. Reproductive health education can improve
knowledge and reduce reproductive health problems among adolescents in developing coun-
tries[28]. Living in a neighborhood with an educated majority expose the mother to women
who are more capable of deciding on appropriate care seeking[29]. An educated woman can
better catch health messages delivered through newspapers, billboards, and other media. Over-
all, formal education challenges traditional beliefs about health and health-seeking and trans-
forms women’s attitudes towards safe delivery[20].
Exposure to media was also a significant predictor of health facility delivery services. Several
studies have reported the effectiveness of media in influencing health service utilisation includ-
ing health facility delivery[30, 31]. Higher concentration of media exposures in the community
also plays an influential role in overall health service utilisation of that community. Increased
media exposure might help to increase discussion of maternal issues within the community.
This finding is similar to a study in Nigeria which reported the mothers residing in communi-
ties with a higher proportion of exposure to electronic and print media had higher odds of
using health facility delivery services [32].
The findings that women who had more contact with antenatal care service have higher
odds of using health facility delivery services might be an indication that such women are bet-
ter informed about the importance of safe delivery from the counselling during antenatal care
attendance. The finding is similar to the results of previous studies done in other countries [23,
24, 33, 34]. Community ANC service utilisation is also a strong predictor of health facility
delivery. Higher community utilisation of ANC services indicates availability or better access
to health facilities in these communities. Also, women attending antenatal care service are
likely to be better informed about the danger of home delivery and could motivate their neigh-
bours who did not participate in antenatal care service.
Place of residence and region were found to be significant predictors of health facility deliv-
ery. The result is consistent with other studies elsewhere [32, 35–37]. The difference in service
utilisation among the urban and rural community as well as the different geographic region
could be due to health service availability, quality of health services as well as access to health
facilities. Our findings indicate that the likelihood of health facility delivery is higher among
older women[20, 38] as well as multiparous women[13, 39, 40]. Earlier studies suggested older
women are better aware of availability and accessibility of such services [20, 38] while multipa-
rous woman develop confidence about childbirth from the experience and knowledge
acquired from their earlier delivery[12, 26, 39]. Finally, similar to the findings of another study
in Bangladesh we found Muslim women to have less probability of using health facility delivery
services than women from other religions [26]. Muslim women may use fewer services at a
health facility due to their conservative behaviour. After adjusting for individual and commu-
nity variables, we found that there was still unexplained variance. Some of this variance might
be explained by other potentially important health system level variables which we did not
have data (e.g., quality of health services, availability of service provider, distance to the nearest
health health facility).
Limitations
There are few limitations in this study. We used cluster—the primary sampling unit of our
research as our definition of a community. However, a cluster has an arbitrary boundary and may
not represent an actual community. We did not include some known predictors of health facility
service utilisation because the data were not available in the survey examined. These included the
availability of health services in the community or distance to the nearest health facility. Also, we
could not include husband education status or decision-making power of the women.
Conclusion
Our findings provide valuable information to the policymakers that can be used when plan-
ning interventions to promote health facility delivery in Bangladesh. In addition to the individ-
ual attributes of women that influenced the use of health facility delivery services, we also
highlighted the community determinants that contributed significantly in health facility deliv-
ery service utilization. Several community level factors significantly predicted the uptake of
health facility delivery care which reinforce the need for community empowerment and focus
on less privileged communities. The evidence suggests the need to go beyond addressing chal-
lenges at individual levels to improve the uptake of facility delivery services. Thus, increasing
the use of health facility delivery services will require strategies that target high-risk groups,
which may be most effectively defined, based on contextual factors such as community pov-
erty, community education status, community exposure to mass media and community use of
other health services. The fact that women education and household wealth are important
determinants for health facility delivery services reinforces the needs for addressing current
disparities in women education and wealth. Promoting intersectoral actions would thus be
vital in improving maternal health. Overall, action is required at all levels–level of the individ-
ual woman and her community.
Acknowledgments
We thank MEASURE Evaluation for granting permission to use the BMMS 2010 data.
Author Contributions
Conceptualization: Tanvir M. Huda.
Data curation: Tanvir M. Huda.
Formal analysis: Tanvir M. Huda.
Methodology: Tanvir M. Huda, Michael J. Dibley.
Resources: Tanvir M. Huda.
Supervision: Michael J. Dibley.
Writing – original draft: Tanvir M. Huda.
Writing – review & editing: Tanvir M. Huda, Morseda Chowdhury, Shams El Arifeen,
Michael J. Dibley.
References
1. World Health Organization. Trends in maternal mortality 1990 to 2015: estimates by WHO, UNICEF,
UNFPA, World Bank Group and the United Nations Population Division. Geneva: WHO; 2015.
2. WHO. Trends in maternal mortality 1990 to 2015: estimates by WHO, UNICEF, UNFPA, World Bank
Group and the United Nations Population Division. Geneva: WHO. 2015.
3. <BDHS1 2014.pdf>.
4. Campbell OM, Graham WJ, Lancet Maternal Survival Series steering g. Strategies for reducing mater-
nal mortality: getting on with what works. Lancet. 2006; 368(9543):1284–99. https://doi.org/10.1016/
S0140-6736(06)69381-1 PMID: 17027735.
5. Koblinsky M, Matthews Z, Hussein J, Mavalankar D, Mridha MK, Anwar I, et al. Going to scale with pro-
fessional skilled care. Lancet. 2006; 368(9544):1377–86. https://doi.org/10.1016/S0140-6736(06)
69382-3 PMID: 17046470.
6. Montagu D, Yamey G, Visconti A, Harding A, Yoong J. Where Do Poor Women in Developing Countries
Give Birth? A Multi-Country Analysis of Demographic and Health Survey Data. PLoS ONE. 2011; 6(2):
e17155. https://doi.org/10.1371/journal.pone.0017155 PMID: 21386886
7. National Institute of Population Research and Training (NIPORT), Mitra and Associates, and ICF Inter-
national. 2016. Bangladesh Demographic and Health Survey 2014. Dhaka, Bangladesh, and Rockville,
Maryland, USA: NIPORT, Mitra and Associates, and ICF International.
8. Parkhurst JO, Rahman SA, Ssengooba F. Overcoming access barriers for facility-based delivery in low-
income settings: insights from Bangladesh and Uganda. J Health Popul Nutr. 2006; 24(4):438–45.
PMID: 17591340; PubMed Central PMCID: PMCPMC3001147.
9. Pervin J, Moran A, Rahman M, Razzaque A, Sibley L, Streatfield PK, et al. Association of antenatal
care with facility delivery and perinatal survival—a population-based study in Bangladesh. BMC Preg-
nancy Childbirth. 2012; 12:111. https://doi.org/10.1186/1471-2393-12-111 PMID: 23066832; PubMed
Central PMCID: PMCPMC3495045.
10. Shahabuddin AS, Delvaux T, Utz B, Bardaji A, De Brouwere V. Determinants and trends in health facil-
ity-based deliveries and caesarean sections among married adolescent girls in Bangladesh. BMJ
Open. 2016; 6(9):e012424. https://doi.org/10.1136/bmjopen-2016-012424 PMID: 27633641; PubMed
Central PMCID: PMCPMC5030621.
11. Enuameh YA, Okawa S, Asante KP, Kikuchi K, Mahama E, Ansah E, et al. Factors Influencing Health
Facility Delivery in Predominantly Rural Communities across the Three Ecological Zones in Ghana: A
Cross-Sectional Study. PLoS One. 2016; 11(3):e0152235. https://doi.org/10.1371/journal.pone.
0152235 PMID: 27031301; PubMed Central PMCID: PMCPMC4816577.
12. Kruk ME, Rockers PC, Mbaruku G, Paczkowski MM, Galea S. Community and health system factors
associated with facility delivery in rural Tanzania: a multilevel analysis. Health Policy. 2010; 97(2–
3):209–16. https://doi.org/10.1016/j.healthpol.2010.05.002 PMID: 20537423.
13. Kruk ME, Hermosilla S, Larson E, Vail D, Chen Q, Mazuguni F, et al. Who is left behind on the road to
universal facility delivery? A cross-sectional multilevel analysis in rural Tanzania. Trop Med Int Health.
2015; 20(8):1057–66. https://doi.org/10.1111/tmi.12518 PMID: 25877211; PubMed Central PMCID:
PMCPMC4490971.
14. Yadav A, Kesarwani R. Effect of Individual and Community Factors on Maternal Health Care Service
Use in India: A Multilevel Approach. J Biosoc Sci. 2016; 48(1):1–19. https://doi.org/10.1017/
S0021932015000048 PMID: 25741587.
15. Leone T, Padmadas Ss Fau—Matthews Z, Matthews Z. Community factors affecting rising caesarean
section rates in developing countries: an analysis of six countries. (0277–9536 (Print)).
16. National Institute of Population Research and Training (NIPORT), MEASURE Evaluation, and Interna-
tional Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b), 2012. Bangladesh Maternal Mor-
tality and Health Care Survey 2010. Dhaka, Bangladesh: NIPORT, MEASURE Evaluation, and icddr,b.
17. Andersen R, Rice T, Kominski G. Changing the US health care system: Key issues in health services,
policy, and management. 2. San Francisco: Jossey-Bass; 2001.
18. Andersen R, Newman JF. Societal and individual determinants of medical care utilization in the United
States. Milbank Mem Fund Q Health Soc. 1973; 51(1):95–124. https://doi.org/10.2307/3349613 PMID:
4198894
19. Andersen R. A behavioral model of families’ use of health services. Chicago: Center for health adminis-
tration studies, University of Chicago. Research Ser; 1968. p. 25.
20. Mezmur M, Navaneetham K, Letamo G, Bariagaber H. Individual, household and contextual factors
associated with skilled delivery care in Ethiopia: Evidence from Ethiopian demographic and health sur-
veys. PLoS One. 2017; 12(9):e0184688. https://doi.org/10.1371/journal.pone.0184688 PMID:
28910341; PubMed Central PMCID: PMCPMC5598994.
21. Ononokpono DN, Odimegwu CO, Imasiku E, Adedini S. Contextual determinants of maternal health
care service utilization in Nigeria. Women Health. 2013; 53(7):647–68. https://doi.org/10.1080/
03630242.2013.826319 PMID: 24093448.
22. Straneo M, Fogliati P, Azzimonti G, Mangi S, Kisika F. Where do the rural poor deliver when high cover-
age of health facility delivery is achieved? Findings from a community and hospital survey in Tanzania.
PLoS One. 2014; 9(12):e113995. https://doi.org/10.1371/journal.pone.0113995 PMID: 25460007;
PubMed Central PMCID: PMCPMC4252065.
23. Colombara DV, Hernandez B, Schaefer A, Zyznieuski N, Bryant MF, Desai SS, et al. Institutional Deliv-
ery and Satisfaction among Indigenous and Poor Women in Guatemala, Mexico, and Panama. PLoS