Journal of Public Health | Vol. 34, No. 2, pp. 261 271 | doi:10.

1093/pubmed/fdr108 | Advance Access Publication 12 January 2012

Frequently used healthcare services in urban slums of Dhaka

and adjacent rural areas and their determinants
M.M.H. Khan1,2, Oliver Grubner3, Alexander Kramer1
1

Department of Public Health Medicine, School of Public Health, University of Bielefeld, Bielefeld, Germany
Department of Statistics, Jahangirnagar University, Savar, Dhaka, Bangladesh
Geomatics Laboratory, Geography Department, Humboldt-Universitat zu Berlin, Berlin, Germany
Address correspondence to MMH Khan, E-mail: mobarak.khan@uni-bielefeld.de
2
3

A B S T R AC T

key determinants of those preferences.

Methods The data were collected through baseline surveys conducted in 2008 and 2009. A total of 3207 subjects aged 10 90 years were
systematically selected from 12 big slums in Dhaka and 3 rural villages outside Dhaka.
Results Two frequently used healthcare sources utilized in 1 month preceding the baseline survey were pharmacies (slum, 42.6%; rural,
30.1%) and government hospitals/clinics (GVHC; slum, 13.5%; rural, 8.9%). According to the multilevel logistic regression analysis adjusted
for age, sex and marital status, the likelihood of using pharmacies and GVHC were higher for those subjects who used non-hygienic toilets,
who reported food deficiency at a family level, who expressed dissatisfaction about family income and who stated poor health status. Some
more factors namely overweight, living in permanently structured house, smoking bidis and less frequency of watching TV were associated
with higher likelihood of using GVHC.
Conclusions Pharmacy was the most dominant healthcare service in both areas. As persons running pharmacies often provide poor quality of
healthcare services, they need continuous training and back-up supports to improve their quality of services and to strengthen the overall
healthcare system in Bangladesh.
Keywords Dhaka, government hospitals, pharmacy, urban slums, utilization of health care

Background
Bangladesh is a signatory of many international treaties and
declarations namely Alma-Ata conference, International
Conference on Population and Development, Beijing
Declaration and the United Nations Millennium
Development Goals.1 This country also has a pluralistic
healthcare infrastructure2 with many options of healthcare
services.3,4 These services could be broadly grouped as
public versus private, or formal versus informal, or allopathic versus non-allopathic or modern versus traditional services.3 5 Ahmed et al.3 proposed ve comprehensive groups
of healthcare options: self-care/self-treatment, pharmacy/
drugstore, traditional treatment, paraprofessional and qualied allopathic practitioner. Although Bangladesh intensied

healthcare services after Alma-Ata conference6 and has

already achieved remarkable progress in health sector,5 unfortunately not everybody had equal access to these services.
Particularly the poorest segment of the society had limited
access to basic healthcare services.7 Such an inequity is, of
course, a violation of human rights1 and an acute challenge
for Bangladesh.7,8
The healthcare system in Bangladesh is generally biased
towards rich people and urban elites.1 Poor people face

M.M.H. Khan, Assistant Professor

Oliver Grubner, Research Associate
Alexander Kramer, Professor and head of Department of Public Health Medicine

# The Author 2012, Published by Oxford University Press on behalf of Faculty of Public Health. All rights reserved.

261

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Background To compare patterns of healthcare service user preference between urban slums in Dhaka and adjacent rural areas and to identify

262

J O U RN A L O F P U B L I C H E A LTH

slum dwellers. Studies comparing healthcare utilization

between urban slum and rural dwellers are also limited in
Bangladesh. To bridge these gaps, rst we aimed to
compare frequently used healthcare services including user
satisfaction between urban slums in Dhaka and adjacent
rural areas. Secondly we aimed to identify some important
determinants for the leading and second leading healthcare
services focusing on slums only. We emphasized slum dwellers for various reasons. One of the reasons was that these
people are rapidly increasing in Dhaka and already sharing
about 40% of the city population. Generally they suffer
more from preventable communicable diseases and poor
quality of life than non-slum afuent populations.32,33
Moreover, these people have limited accessibility to the basic
healthcare services because of their poor socio-economic
conditions and other barriers.
The present study might be interesting for health planners
and other relevant stakeholders in developing countries as
they need to understand peoples healthcare utilization
pattern including determinants of leading healthcare services. This information could facilitate formulating policies
and implementing targeted strategies which are responsive
to peoples needs and priorities.3,28

Methods
The data were collected using two baseline surveys (designed
for 1-year cohort study) conducted between March and
April in the year of 2008 and 2009. This study was conducted under the priority programme of German Research
Foundation Megacities Megachallenge: Informal Dynamics
of Global Change. As the major focus of the programme
was to study health outcomes among slums dwellers in
Dhaka, we included more samples from slum areas than
rural areas. In 2008, we interviewed a total of 1269 adults,
662 adults from 3 urban slums and 607 adults from 3 rural
villages. Later in 2009 we increased our sample size by conducting a similar survey with 1938 adults living in 9 additional slums in Dhaka. Finally, our total sample was 3207
respondents. In both surveys, we used an almost similar
questionnaire which was pre-tested before collecting the
data.
Sampling strategy

We used a systematic sampling approach for selecting slums

and households in Dhaka (Fig. 1). Firstly, we used secondary
information regarding slums in Dhaka (e.g. name of the
slum, number of households, estimated population and location of the slum) provided by the Centre for Urban Studies

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

more barriers to healthcare services than the rich.9 Some of

the barriers are acute shortage of healthcare workforce,5,10
poor quality of services, absence of critical staff, lack of essential drugs and supplies, lack of supervision and job accountability.7 The healthcare inequity is not only unfair but
also challenging for many developing countries because it
puts already disadvantaged people at further disadvantages
and undermines population health and retards overall development.6,11 18 Particularly inadequate access to healthcare
services increases health risks for poor people (e.g. slum
dwellers) suffering from more infectious diseases and other
illnesses for a prolonged period of time.18
Since Bangladesh is committed to provide basic healthcare services to its entire population, ofcially many services
at public healthcare facilities are free or available with a
minimum fee. Unfortunately, these services are still underutilized in Bangladesh for various reasons, some of them are
already mentioned above. According to our knowledge, in
Bangladesh like other neighbouring countries pharmacies
often cluster around government hospitals and doctors refer
patients to them with prescriptions either out of commercial
considerations or they have a deal with the owners or they
are owners. Another important aspect that undermines the
utilization of public healthcare services is widely practised
corruption (associated with higher informal and unofcial
charging) at all levels.2,19 Moreover, public healthcare services are commonly perceived as poor quality services
because of inadequate drug supplies and insufcient management.2,8 All these problems in public healthcare facilities
facilitate private health care to grow. Consequently various
types of private healthcare facilities including pharmacies are
mushrooming everywhere and dominating the total healthcare system in Bangladesh.2,10,20 Unfortunately, private
healthcare services are also poorly regulated and rarely registered by the concerned authorities of the government.2
Within the domain of private healthcare sector, pharmacies (mostly run by drug sellers) are the leading source of
care for common and acute medical problems in
Bangladesh20,21 and other developing countries.22 25 Some
possible reasons for utilizing pharmacies may include close
proximity, short distance, short waiting time, easy accessibility and reduced cost for treatment.4,10,22,26,27 Other determinants may include socio-demographic and economic factors,
cultural beliefs and practices, severity of diseases, political
and healthcare system.28 30 All these determinants could
also be grouped as predisposing, enabling and need
factors.31
Although many studies regarding utilization and customers satisfaction of healthcare services are available, such
information is scarce in Bangladesh particularly for urban

FR EQUE N TLY US ED HE ALTHCA R E SERV I C ES I N UR BA N SLUMS I N D H AK A

Analysis of slum information provided by CUS and

selection of big slums

Mapping households/families in the selected slums

Providing unique identification number for each

household to make a sampling frame
Calculation of sample size for each slum using a
statistical formula
Used systematic sampling (rth household) approach to
locate household for interview

263

baseline survey. We also used global positioning system to

record the location of each interviewed household. Although
rural villages were selected purposively, we applied similar
techniques from mapping to data collection (as shown
in Fig. 1).
Information was collected through pre-tested questionnaire. Face-to-face interviews by trained university graduates
were performed in the day time. Before starting the interview, the aims of the survey were explained and verbal
consent from each respondent was taken. Participation was
voluntary and freedom was given to quit the interview
anytime. Only some relevant variables were used in this
study.

Fig. 1 Sampling scheme for selecting slums and collecting data in Dhaka.

(CUS) in 2005.34 According to the list of CUS, there were

4900 slums in Dhaka. Secondly, to select only big slums,
we used two inclusion criteria: a minimum number of 500
households in the slum and a minimum area of 6 acres land.
To increase the representativeness of the study area, we purposively selected some administrative units from the city that
were not very adjacent in terms of geographical location. In
units with two or more slums, we randomly selected one
slum. It should be noted that some listed slums were evicted
or turned into afuent residential areas or open spaces during
our eld study. Thirdly, we prepared individual household
map (taking Google maps as a base) for each of the selected
slums. In the map, every household was given a unique identication number. Fourthly, we estimated the representative
sample of households/families for each slum using a statistical formula (not given here) proposed by Bartlett et al.35 In
our study, we used a 95% condence level (i.e. alpha 0.05)
and an acceptable error margin of d 6%. Since it was not
possible to conduct a pilot study for estimating prevalences
of various outcomes of interest (P), we choose the recommended value of P 0.50, which can provide a maximum
variance and a maximum sample size. Our samples varied
from slum to slum depending on the number of households
in slums. Slums with highest and smallest number of households required highest and smallest number of samples,
respectively. Fifthly, we calculated the sampling rate r by
dividing the number of families in the slum by the calculated
sample size. We then interviewed an adult in every rth
household. We did not adjust our sample size for the nonresponse rate as we achieved our target by replacing the nonrespondent household by next available household at a

Outcome variables

Our outcomes of interest were utilization of healthcare services by the respondents. Every respondent was asked to
answer (yes/no) of the following question:
During the last one month preceding the survey, did you
visit any health facility for healthcare?
When the answer was yes, respondent was asked to
mention the name of the healthcare sources. There were 20
different options in the questionnaire (multiple answers were
recorded). Considering a small number of frequencies and
similarities of some healthcare services, we made seven
broad categories. These categories are somehow consistent
with categories of other studies.3,4
(i)
(ii)
(iii)
(iv)

Pharmacies/drug stores/sellers (PHAR)

GVHC
Private hospitals/clinics (PVHC)
Qualied allopathic practitioners (QUALP): MBBS/
trained doctors
(v) Paraprofessionals (PARAP) which include government health worker, NGO health worker, family
planning worker and nurse/trained TBA and providers working at satellite clinic, family health welfare
centre, community clinic, maternity clinic and NGO
clinic.
(vi) Traditional health providers/practitioners (TRADP):
ayurvedic/unani, homeopathic and Kabiraj
(vii) Others (OTHER).
All categories except PHAR combined two or more sources.
We dichotomized all combined categories for analyses and
assigned a value of 1 when respondent visited at least one
source belonged to that category and 0 when visited none
of them. Not all categories were equally analysed. We only

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Selection of adults for interview and administration of

pre-tested questionnaire for data collection

264

J O U RN A L O F P U B L I C H E A LTH

considered PHAR and GVHC for detailed analyses because

they were the two dominant sources in the total sample.
Independent variables

Multilevel analysis

A multilevel analysis is a recent statistical method used to

link explanatory variables from different levels without committing ecological and atomistic fallacy.36 38 Our data were
hierarchical with levels. The units at a lower level were
respondents who were nested within a higher level where
areas were higher level units. Due to the nested structure of
the data, respondents from the same area may share
common exposures and hence the odds of outcomes (e.g.
respondents seeking health care from PHAR or GVHC)
were not independent. A traditional logistic regression
(which is single level in multilevel terms) assumes: (a) independence of the observations conditional on the explanatory
variables and (b) uncorrelated residual errors. These
assumptions are not always met when analysing the nested
data.38 Considering these limitations, we applied multilevel
logistic regression which allows simultaneous examination of
the effects of area/group level and individual level variables
on individual level outcomes while accounting for the nonindependence of observations within groups.36,38,39

We performed various statistical analyses starting from

simple frequency, cross-table to multilevel binary logistic regression. A frequency analysis was performed to provide descriptive information (e.g. mean and percentage) about the
study population. Cross-table analyses were performed to
check the bivariate association of each outcome variable
with each independent variable. Finally multilevel logistic regression was performed to nd signicant variables associated with both outcome variables. Most of the analyses
were performed through SPSS 17.0. Only multilevel modelling for two outcome variables ( pharmacy and GVHC) was
performed through MLwiN. Data used in MLwiN were
generated in SPSS. Satisfaction levels regarding healthcare
sources were also reported.

Results
Comparison of background information by urban
slums and rural area

Table 1 presents background information of the respondents. Slum respondents were younger than their rural counterparts (P , 0.001). Family size was higher (P , 0.001) and
educational level was better among rural respondents (P ,
0.001). Smoking bidis was signicantly higher in the rural
area. In contrast, smoking cigarettes was higher in urban
slums. Tap water was the leading source of water in urban
slums, whereas it was tubewell in rural areas. The most
common type of toilet was slab/pit latrine in both areas.
Self-reported health status was better in rural than urban
areas (P , 0.001).
Table 2 is used to provide comparative information about
utilization of healthcare sources. For the total sample, major
sources of health care were pharmacies, followed by GVHC,
QUALP and PVHC. The use of GVHC and pharmacies
were signicantly higher among slum dwellers. Other
sources namely PVHC, QUALP and TRADP were signicantly higher among rural inhabitants. Most of the respondents utilized only one healthcare source (56.3%). About 50
and 38% visited either pharmacy or GVHC in urban slums
and rural areas respectively.
Table 3 compares self-reported satisfaction about different healthcare sources by urban slums and rural areas.
Satisfaction levels differed signicantly for two sources
namely PVHC (P 0.007) and pharmacies (P , 0.001).
Satisfaction level of PVHC was signicantly higher in slums,
whereas it was signicantly lower for pharmacies.
Healthcare utilizations for common health problems were
also different between urban slums and rural area (Table 4).

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Only a few variables namely age (10 30, 31 50 and 51

years), sex (male and female), marital status (unmarried,
married, divorced/separated/widowed), education (no education, 1 5 years education and 6 years education), body
mass index (BMI) calculated using self-reported height and
weight (normal weight BMI from 18.5 to 24.99; underweight BMI, 18.5 and overweight BMI .25.00), types of
house ( permanent and provisional/mess), frequency of
watching TV (0 3 and 4 7 days/week), smoking cigarettes
(no or yes), smoking bidis (no and yes), types of toilet (hygienic: modern/septic/pit/slab latrine; non-hygienic: open/
hanging/else), whether family got sufcient food (no or
yes), whether respondent was satised by the family income
(satised, moderately satised and dissatised), and selfperceived health status (good, medium and poor) and study
sites/area (15 sites) were used as covariates. Out of them,
three variables namely age, sex and marital status were
adjusted in a multilevel analysis as they were signicantly
associated with any of the two outcomes (not shown).
Adjusted odds ratio (OR) and 95% condence intervals
(CIs) for all the remaining variables are presented in
Table 5. Study sites/area was used as a second level variable in a multilevel logistic regression analysis (only for
slums).

Statistical analysis

FR EQUE N TLY US ED HE ALTHCA R E SERV I C ES I N UR BA N SLUMS I N D H AK A

265

Table 1 Comparison of background characteristics of the respondents

Background characteristics

Urban slums
n

Rural area
%a

P
%a

Age in years (mean + SD)

2600

34.61 + 13.15

607

37.21 + 13.13

0.000

Family size (mean + SD)

2600

4.29 + 1.63

607

4.95 + 1.70

0.000

Monthly family income in Dhaka (mean + SD)b

2598

6684.0 + 4642.5

580

6462.8 + 4638.1

0.299

Head of the family (male)

2340

90.0

575

94.7

0.000

Sex (male)

1301

50.1

312

51.4

0.546
0.000

Education
No education

1648

63.4

194

32.0

1 5 years education

562

21.6

164

27.0

6 years education

390

15.0

249

41.0

BMIb
1553

59.7

396

65.5

Underweight

767

29.5

157

25.9

Overweight

267

10.3

54

8.9

Never married

182

7.0

63

10.4

Married

2309

88.8

538

88.6

Divorced/separated/widowed

109

4.2

1.0

0.060

Marital status
0.000

Smoking history
Smoking bidi (yes)

329

12.7

127

20.9

0.000

Smoking cigarette (yes)

666

25.6

90

14.8

0.000

0 3 days/week

1614

62.1

372

61.3

0.718

4 7 days/week

986

37.9

235

38.7

Frequency of watching TV

Sources of drinking water

Tape water

1498

57.6

92

15.2

Tubewell

901

34.7

523

84.5

Else

201

7.7

0.3

Modern toilet

502

19.3

28

4.6

Slab latrine

1371

52.7

435

71.1

Open/hanging

727

28.0

144

23.7

Good

566

21.8

248

40.9

Medium

1456

56.0

304

50.1

Poor

578

22.2

55

9.1

0.000

Types of toilet
0.000

Self-reported health status

0.000

All results are expressed as percentages unless otherwise stated.

Variables had some missing values.

Pharmacies were the most dominant source of health care

for all common problems in both areas. GVHC were the
second most dominant source of health care in slum areas
but not in rural areas. Visiting qualied practitioners (i.e.
MBBS) was the second most important option in rural
areas. As compared with pharmacies, all other options were
not frequently used.

Results of multilevel analysis for urban slums

According to the multilevel null model with random intercept, the variances of the intercepts (as compared with their
corresponding standard errors) were 2.1 (0.124/0.059) for
pharmacies and 1.3 (0.049/0.037) for GVHC (Table 5).
According to the guidelines of Twisk,40 such a null model is
only meaningful for PHAR because the variance was more

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Normal weight

266

J O U RN A L O F P U B L I C H E A LTH

Table 2 Comparison of healthcare utilization by urban slums and rural area

Urban slums (total n 2600)

Rural area (total n 607)

Visited

Visited

PHAR

1107

42.6

183

30.1

GVHC

352

13.5

54

8.9

0.002

QUALP

70

2.7

98

16.1

0.000

PVHC

101

3.9

51

8.4

0.000

PARAP

66

2.5

10

1.6

0.194

TRADP

36

1.4

19

3.1

0.003

OTHER

0.0

10

1.6

0.000

Sources

Sources visited
0.000

Number of sources visited

1004

38.6

225

37.1

Visited only one source

1464

56.3

342

56.3

132

5.1

40

6.6

Visited two or more sources

0.300

Pharmacy/GVHC users
Visited neither PHAR nor GVHC

1216

46.8

374

61.6

Visited either PHAR or GVHC

1309

50.3

229

37.7

Visited both PHAR and GVHC

75

2.9

0.7

0.000

Table 3 Comparison of satisfaction about healthcare sources by urban slums and rural area
Name of sources visited

Satisfaction about source

Dissatisfied

Moderately satisfied

Satisfied
n

PHAR

127

11.5

614

55.5

366

33.1

0.000

GVHC

87

24.7

175

49.7

90

25.6

0.069

QUALP

14

20.0

34

48.6

22

31.4

0.102

PVHC

15

14.9

44

43.6

42

41.6

0.007

PARAPs

10.6

38

57.6

21

31.8

0.590

TRADP

22.2

19

52.8

25.0

0.287

OTHER

0.0

100.0

0.0

0.231

49.2

Urban slums

Rural area
PHAR

36

19.7

57

31.1

90

GVHC

16

29.6

18

33.3

20

37.0

QUALP

34

34.7

36

36.7

28

28.6

PVHC

19

37.3

17

33.3

15

29.4

PARAP

20.0

60.0

20.0

TRADP

26.3

31.6

42.1

OTHER

70.0

20.0

10.0

than two times higher than its own standard error.40

Although the null model with random intercept was not
meaningful for GVHC, we applied multilevel techniques for
both outcomes.

Table 5 also presents the multilevel results for different

variables adjusted for age, sex and marital status.
Overweight respondents visited GVHC more as compared
with respondents with normal weight (OR 1.43; 95%

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Visited no source

FR EQUE N TLY US ED HE ALTHCA R E SERV I C ES I N UR BA N SLUMS I N D H AK A

267

Table 4 Self-reported common health problems and utilization of specific healthcare services
Health problems

Visited healthcare facility (yes)

PHAR

GVHC

QUALP

PVHC

PARAP

TRADP

OTHER

n (%)

n (%)

n (%)

n (%)

n (%)

n (%)

n (%)

655 (61.1)

121 (11.3)

28 (2.6)

37 (3.5)

26 (2.4)

15 (1.4)

1 (0.1)

76 (69.1)

10 (9.1)

16 (14.5)

9 (8.2)

3 (2.7)

3 (2.7)

1 (0.9)

315 (58.8)

65 (12.1)

12 (2.2)

20 (3.7)

11 (2.1)

7 (1.3)

0 (0.0)

25 (40.3)

6 (9.7)

18 (29.0)

5 (8.1)

1 (1.6)

3 (4.8)

1 (1.6)

127 (54.3)

43 (18.4)

3 (1.3)

13 (5.6)

10 (4.3)

5 (2.1)

0 (0.0)

42 (46.2)

21 (23.1)

27 (29.7)

6 (6.6)

2 (2.2)

4 (4.4)

1 (1.1)

117 (49.8)

36 (15.3)

10 (4.3)

14 (6.0)

9 (3.8)

3 (1.3)

0 (0.0)

13 (50.0)

4 (15.4)

2 (7.7)

2 (7.7)

0 (0.0)

Slums (N 172)

72 (41.9)

26 (15.1)

9 (5.2)

8 (4.7)

5 (2.9)

0 (0.0)

Rural area (N 21)

11 (52.4)

3 (14.3)

5 (23.8)

3 (14.3)

2 (9.5)

0 (0.0)

0 (0.0)

48 (45.3)

11 (10.4)

6 (5.7)

6 (5.7)

4 (3.8)

0 (0.0)

0 (0.0)

8 (34.8)

1 (4.3)

8 (34.8)

7 (30.4)

0 (0.0)

2 (8.7)

0 (0.0)

67 (55.8)

25 (20.8)

2 (1.7)

3 (2.5)

2 (1.7)

1 (0.8)

0 (0.0)

4 (50.0)

1 (12.5)

2 (25.0)

0 (0.0)

1 (12.5)

0 (0.0)

1 (12.5)

Fever
Slums (N 1072)
Rural area (N 110)
Cold/cough
Slums (N 536)
Rural area (N 62)
Gastric
Slums (N 234)
Rural area (N 91)
Slums (N 235)
Rural area (N 26)

8 (30.8)

7 (26.9)

Weakness
14 (8.1)

Body pain
Slums (N 106)
Rural area (N 23)
Diarrhoea
Slums (N 120)
Rural area (N 8)

N means total respondents experienced specific health problem and n means samples who visited a specific healthcare source for a specific health
problem

CI 1.00 2.05) but they visited PHAR less (OR 0.67;

95% CI 0.51 0.89). People living in a permanently structured house showed a higher likelihood of visiting GVHC
as compared with those in provisional/mess house. People
using non-hygienic toilets (hanging or open latrine) showed
consistently higher likelihoods of visiting GVHC (OR
1.69; 95% CI 1.31 2.19). Respondents with insufcient
foods for family reported signicantly higher likelihood of
visiting both healthcare sources. Compared to respondents
who reported satisfaction about family income, other two
groups (moderately satised and dissatised) visited both
healthcare sources more. Similarly respondents with moderate and poor self-perceived health visited both sources more
as compared with respondents with good health.

Discussion
Main findings of the study

Many healthcare provisions are available for urban slum and

rural areas in Bangladesh. But these provisions were utilized

differently between these areas. Pharmacies were the most

dominant source of healthcare in both areas for most of the
common health problems. The other main sources were
GVHS, MBBS and PVHC, respectively. According to our
study, GVHC was underutilized although many services are
generally free or less costly there. Visitors of GVHC facilities reported lowest percentage of satisfaction, which might
be associated with long waiting time.41 Other factors of
lower customer satisfaction may include inconvenient hours,
unavailability of drugs, corruptions and the unfriendly
attitude of the service providers.42

What is already known on this topic?

Health services inequality is common in all developing

countries including Bangladesh. Accessibility to and use of
healthcare services are limited for the poor segment of
the society. The overall quality of the healthcare services
is poor in developing countries. Corruptions, mainly in
the form of higher informal and unofcial charging, in

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

Headache

268

J O U RN A L O F P U B L I C H E A LTH

Table 5 Results of multilevel logistic regression analyses related to binary outcomes reflecting sources of healthcare (only for slums)
Variables

GVHC
Coefficient (SE)

Area-level variance (null model)

PHAR
OR

95% CI

0.049 (0.037)

Coefficient (SE)

OR

95% CI

0.124 (0.059)

Education (edu.)
1 5 years edu./no edu.

20.164 (0.154)

0.85

0.63 1.15

20.067 (0.103)

0.94

0.76 1.14

6 years edu./no edu.

20.023 (0.180)

0.98

0.69 1.39

20.193 (0.124)

0.82

0.65 1.05

Underweight/normal

0.051 (0.133)

1.05

0.81 1.37

0.123 (0.092)

1.13

0.94 1.35

Overweight/normal

0.360 (0.183)

1.43

1.00 2.05

20.396 (0.144)

0.67

0.51 0.89

0.670 (0.178)

1.95

1.38 2.77

0.010 (0.151)

1.01

0.75 1.36

0.209 (0.125)

1.23

0.96 1.57

0.093 (0.085)

1.10

0.93 1.30

20.339 (0.162)

0.71

0.52 0.98

0.032 (0.105)

1.03

0.84 1.27

0.420 (0.175)

1.52

1.08 2.14

0.111 (0.129)

1.12

0.87 1.44

0.526 (0.132)

1.69

1.31 2.19

0.124 (0.102)

1.13

0.93 1.38

0.390 (0.125)

1.48

1.16 1.89

0.273 (0.085)

1.31

1.11 1.55

Moderately satisfied/satisfied

0.468 (0.230)

1.60

1.02 2.51

0.577 (0.144)

1.78

1.34 2.36

Dissatisfied/satisfied

0.706 (0.227)

2.03

1.30 3.16

0.826 (0.143)

2.28

1.73 3.02

Moderate/good

0.636 (0.176)

1.89

1.34 2.67

0.896 (0.111)

2.45

1.97 3.05

Poor/good

0.765 (0.198)

2.15

1.46 3.17

0.667 (0.134)

1.95

1.50 2.53

BMI

Types of house
Permanent/non-permanent
Watching TV

Yes/no
Smoking bidi
Yes/no
Types of toilet
Non-hygienic/hygienic
Family get sufficient food
No/yes
Satisfied by family income

Self-perceived health status

Note: adjusted for age (three groups), sex and marital status (three groups).

public healthcare services are also widely practised in

Bangladesh.2,19 Private healthcare services are mushrooming in developing countries without proper registration
and supervision. Particularly pharmacies become the
leading source of healthcare services for majority of poor
people, although the overall quality of such services is low.
Our results were consistent with the ndings of other
studies in Bangladesh20,21 and developing countries.22 25
Although .40% of the pharmacy users were not satised
by the services, still they like to visit pharmacies for their
health problems. This nding was also consistent with the
nding of another study in Bangladesh.20 Some possible
reasons to visit pharmacies are close proximity and easy
accessibility, short distance because of wide availability and
reduced treatment cost.4,10,22,26,27 Sometimes less severity
of diseases inuences the utilization of healthcare services
from pharmacies.27

Healthcare services in terms of counselling, diagnosis and

treatment are often poor in pharmacies.21,23,25,43 For instance, only 8% of the drug dispensers in Dhaka (working
in pharmacies) correctly treated dysentery patients (a
common disease), whereas this gure was 44% for doctors
with postgraduate training.21 Poor referral system and irrational selling of drugs including antibiotics by drug sellers
in pharmacies also limit the quality of healthcare services.23,26,27 Prescribing drugs by untrained and/or unregulated pharmacy staff is not only irrational and harmful, but
also a waste of scarce resources.22,43 Pharmacies often sell
drugs on demand without a prescription. Drug sellers in
pharmacies are minimally educated with little professional
training.22,25 It should be noted that the irrational use of
drugs is common in Bangladesh and elsewhere,21,25 which
may be inuenced by staff education, customer demand,
regulations and economic incentives.43

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

3 4 days/4 7 days per week

Smoking cigarette

FR EQUE N TLY US ED HE ALTHCA R E SERV I C ES I N UR BA N SLUMS I N D H AK A

What this study adds?

investigated through further studies to revise policies and redene their role, growth and coverage and to develop appropriate interventions so that they become more quality
focused and user friendly.7 Monitoring and time-to-time
evaluation of healthcare provision,44 assessing expectations
of healthcare professionals, ensuring standards of ethical
and pharmacy practice through regulation, enforcement
mechanisms and monitoring are important.24,43 Particularly,
proper motivation from higher authority to apply retained
knowledge of drug sellers in prescribing reasonable and
rational drugs for effective management is extremely necessary.23 Education of patients may also play an important
part in reducing the use of potentially dangerous drugs by
enhancing an individuals capacity to assess services on
offer, to judge a providers competence and to evaluate
whether costs are justied and reasonable.3 As the recipients
of health care can provide valuable information to revise the
health system, their opinions and expectations should also
be studied thoroughly.7,24 The last but not least way to
improve the quality of services is to control corruption from
all healthcare sectors.
Limitations of the study

The limitations of the study include the cross-sectional

nature of the study which precludes comment on cause
effect relationships. Findings based on the sample of only
one city may limit the generalizability of the results.
Self-reported information (regarding, e.g. BMI) including
recall bias may also limit the reliability of our ndings.
Moreover, the use of word pharmacy as drug sellers may
be misleading as there are few pharmacies with a qualied
pharmacist who prescribes medicines.26 However, our
limited data did not allow us to distinguish the level of pharmacies based on e.g. education and professional training of
the persons running pharmacies. Finally, the multilevel
analysis may not be useful for GVHC because of its low
intraclass correlation coefcient (1.5%).

Conclusions
Pharmacies were the most dominant healthcare source in
both urban slums of Dhaka and rural areas. Although many
other healthcare provisions including low-cost government
services are available, these were not widely used. Proper
strategies should be developed to reduce existing barriers for
improving overall healthcare services. Particularly persons
running pharmacies need to be regularly monitored by the
relevant authorities. Professional training and back-up supports are also needed for them to improve the quality of

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

The present study is an important contribution to the

limited research focusing on health-seeking behaviours by
urban slum and rural dwellers. We mainly focused on the
health-seeking behaviours of slum dwellers which are not
adequately addressed until recently. Comparative studies
between urban slums and rural areas are also interesting
because slum dwellers are remarkably different from their
rural counterparts in terms of education, smoking, drinking
water, toilet and health seeking.33 For instance, receiving
healthcare services from the QUALP are more common in
rural areas than urban slums. Satisfaction level about different healthcare services also depends on area of living. We
provided detailed information regarding healthcare utilization based on common health problems. We also identied
several determinants for the utilization of pharmacies and
GVHC using the multilevel logistic regression model which
is a more appropriate technique than the traditional analysis.
These ndings might help to develop strategies for improving healthcare qualities of the major sources.
In a developing country such as Bangladesh, pharmacies
are often seen as a rst point of contact for most of the
common health problems.22,25 As compared with any other
healthcare facilities, they get more opportunities to interact
with and advise patients.25 As they have already become an
integral part of the present healthcare system, their service
quality must be improved. Particularly, they can improve the
healthcare for the population belonging to the disadvantaged sections of the society who do not have enough
resources to visit, e.g. private clinics.24 Here we propose
some strategies which might be applicable to pharmacies or
other healthcare sources. With respect to the whole healthcare system, a holistic approach that will take into account
the systems and organizations of health care including regulation, training and education, processes and operation in
the provision and delivery of care25 is needed to improve
the treatment quality for common illnesses, to meet the
changing needs of modern medicine users and to reduce irrational drug dispensing.24 Knowledge and practice of drug
dispensers regarding management and pathogenesis should
be improved for common illnesses.21 The barriers to the
provision of higher quality care and the ways to overcome
them should be identied.25
Basic training programmes should be made available to
the full spectrum of health-care providers who are of greatest importance to the poor, and managerial and regulatory
measures should be enforced to control the use of potentially dangerous drugs.22,26 Service and treatment quality of
the major healthcare providers should be thoroughly

269

270

J O U RN A L O F P U B L I C H E A LTH

services as well as to strengthen the existing healthcare system

in Bangladesh.

15 Jankovic J, Simic S, Marinkovic J. Inequalities that hurt: demographic, socio-economic and health status inequalities in the utilization of
health services in Serbia. Eur J Public Health 2009;1 8.
doi:10.1093/eurpub/ckp189

Funding

16 Mohanty SK, Pathak PK. Rich-poor gap in utilization of reproductive and child health services in India, 1992 2005. J Biosoc Sci
2009;41:381 98.

The authors would like to thank the German Research

Foundation (DFG) for funding the research project
INNOVATE under its priority programme 1233
Megacities-Megachallenge - Informal Dynamics of Global
Change (KR 947/9-1,2,3).

References

2 Vaughan JP, Karim E, Buse K. Healthcare system in transition III.

Bangladesh, Part I. An overview of the healthcare system in
Bangladesh. J Public Health Med 2000;22:5 9.
3 Ahmed SM, Tomson G, Petzold M et al. Socioeconomic status
overrides age and gender in determining health-seeking behaviour
in rural Bangladesh. Bull World Health Organ 2005;83:109 17.
4 Ahmed SM, Hossain MA, Chowdhury MR. Informal sector providers in Bangladesh: how equipped are they to provide rational care?
Health Policy Plan 2009;24:467 78.
5 Anwar I. Bangladesh Health System in Transition: Challenges and
Opportunities. Dhaka: Institute of International Public Health, Darul
Ihsan University, 2010.
6 Razzaque A, Streateld PK, Gwatkin DR. Does health intervention
improve socioeconomic inequalities of neonatal, infant and child
mortality? Evidence from Matlab, Bangladesh. Int J Equity Health
2007;6:4. doi:10.1186/1475-9276-6-4
7 Andaleeb SA. Public and private hospitals in Bangladesh: service
quality and predictors of hospital choice. Health Policy Plan
2000;15(1):95 102.
8 Osman FA. Health policy, programmes and system in Bangladesh:
achievements and challenges. South Asian Sur 2008;15:263 88.
9 Chowdhury ME, Ronsmans C, Killewo J et al. Equity in use of
home-based or facility-based skilled obstetric care in rural
Bangladesh: an observational study. Lancet 2006;367:327 32.
10 Mahmood SS, Iqbal M, Hani SMA et al. Are village doctors in
Bangladesh a curse or a blessing? BMC Int Health Hum Rights
2010;10:18.
11 Ahmed SM, Adams AM, Chowdhury M et al. Gender, socioeconomic development and health-seeking behaviour in
Bangladesh. Soc Sci Med 2000;51:361 71.
12 Wagstaff A. Poverty and health sector inequalities. Bull World Health
Organ 2002;80:97 105.
13 Karim F, Tripura A, Gani MS et al. Poverty status and health equity:
evidence from rural Bangladesh. Public Health 2006;120:193 205.
14 Bhuiya A, Hani SMA, Urni F et al. Three methods to monitor utilization of healthcare services by the poor. Int J Equity Health
2009;8:29. doi:10.1186/1475-9276-8-29

18 Kumar A, Mohanty SK. Intra-urban differentials in the utilisation

of reproductive healthcare in India, 1992 2006. J Urban Health
2011;88:311 28.
19 Killingsworth JR, Hossain N, Hendrick-Wong Y et al. Unofcial
fees in Bangladesh: price, equity and institutional issues. Health
Policy Plan 1999;14:152 63.
20 Larson CP, Saha UR, Islam R et al. Childhood diarrhoea management practices in Bangladesh: private sector dominance and continued inequities in care. Int J Epidemiol 2006;35:1430 9.
21 Ronsmans C, Islam T, Bennish ML. Medical practitioners knowledge of dysentery treatment in Bangladesh. BMJ 1996;313:205 6.
22 Butt ZA, Gilani AH, Nanan D et al. Quality of pharmacies in
Pakistan: a cross-sectional survey. Int J Quality Health Care
2005;17:307 13.
23 Khan MMH, Wolter S, Mori M. Post-training quality of syndromic
management of sexually transmitted infections by chemists and
druggists in Pokhara, Nepal: is it satisfactory? Int J Quality Health
Care 2006;18:66 72.
24 Basak SC, van Mil JWF, Sathyanarayana D. The changing roles of
pharmacists in community pharmacies: perception of reality in
India. Pharm World Sci 2009;31:612 8.
25 Smith F. The quality of private pharmacy services in low and
middle-income countries: a systematic review. Pharm World Sci
2009;31:351 61.
26 Kamat VR, Nichter M. Pharmacies, self-medication and pharmaceutical marketing in Bombay, India. Soc Sci Med 1998;47:779 94.
27 Wazaify M, Al-Bsoul-Younes A, Abu-Gharbieh E et al. Social perspectives on the role of community pharmacists and
over-the-counter drugs in Jordan. Pharm World Sci 2008;30:884 91.
28 Shaikh BT, Hatcher J. Health seeking behaviour and health service
utilisation in Pakistan: challenging the policy makers. J Public Health
2004;27:49 54.
29 Taffa N, Chepngeno G. Determinats of health care seeking for
childhood illnesses in Nairobi slums. Trop Med Int Health
2005;10:240 5.
30 Sreeramareddy CT, Shankar RP, Sreekumaran BV et al. Care seeking
behaviour for childhood illnessa questionnaire survey in western
Nepal. BMC Int Health Hum Rights 2006;6:7. doi:10.1186/
1472-698X-6-7
31 Chakraborty N, Islam MA, Chowdhury RI et al. Determinants of
the use of maternal health services in rural Bangladesh. Health
Promot Int 2003;18(4):327 37.
32 Khan MMH, Kraemer A. Socio-economic factors explain differences in public health-related variables among women in

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

1 Rahman RM. Human rights, health and the state in Bangladesh.

BMC Int Health Hum Rights 2006;6:4. doi:10.1186/1472-698X-6-4

17 Victora CG, Matijasevich A, Silveira MF et al. Socio-economic and

ethnic group inequities in antenatal care quality in the public and
private sector in Brazil. Health Policy Plan 2010;25:253 61.

FR EQUE N TLY US ED HE ALTHCA R E SERV I C ES I N UR BA N SLUMS I N D H AK A

271

Bangladesh: a cross-sectional study. BMC Public Health 2008;8:254.

doi:10.1186/1471-2458-8-254

38 Khan MHR, Shaw EH. Multilevel logistic regression analysis applied

to binary contraceptive prevalence data. J Data Sci 2011;9:93 110.

33 Khan MMH, Kraemer A, Gruebner O. Comparison of

health-related outcomes between urban slums, urban afuent and
rural areas in and around Dhaka megacity, Bangladesh. Die Erde
2009;140:69 87.

39 Diez-Roux AV. Multilevel analysis in public health research. Annu

Rev Public Health 2000;21:171 92.

34 Centre for Urban Studies (CUS), National Institute of Population

Research and Training (NIPORT) and Measure Evaluation. Slums
of urban Bangladesh: mapping and census, 2005. Dhaka,
Bangladesh; Chapell Hill, USA, 2006.

41 Rao KD, Peters DH, Bandeen-Roche K. Towards patient-centered

health services in Indiaa scale to measure patient perceptions of
quality. Int J Quality Health Care 2006;18:414 21.

35 Bartlett JE, Kotrlik JW, Higgins CC. Organizational research: determining appropriate sample size in survey research. Informat Technol
Learn Perform J 2001;19:43 50.
36 Hofoss D, Veenstra M, Krogstad U. Multilevel analysis in health research: a tutorial. Ann Ist Super Sanita 2003;39(2):213 22.

42 Bahari MB, Ling YW. Factors contributing to customer satisfaction with

community pharmacies in Malaysia. J Public Health. 2010;18:3541.
43 Chuc NTK, Larsson M, Do NT et al. Improving private pharmacy
practice: a multi-intervention experiment in Hanoi, Vietnam. J Clin
Epidemiol 2002;55:1148 55.
44 Jenkinson C, Coulter A, Bruster S et al. Patients experiences and
satisfaction with health care: results of a questionnaire study of specic aspects of care. Qual Saf Health Care 2002;11:335 9.

Downloaded from http://jpubhealth.oxfordjournals.org/ by guest on March 13, 2015

37 Leyland AH, Groenewegen PP. Multilevel modelling and public

health policy. Scand J Public Health 2003;31:267 74.

40 Twisk JWR. Applied Multilevel Analysis: A Practical Guide. Cambridge:

Cambridge University Press, 2006.