Pan et al.

BMC Family Practice (2016) 17:148

DOI 10.1186/s12875-016-0547-3

RESEARCH ARTICLE Open Access

Knowledge, attitudes and practices towards

antibiotic use in upper respiratory tract
infections among patients seeking primary
health care in Singapore
Darius Shaw Teng Pan1†, Joyce Huixin Huang1†, Magdalene Hui Min Lee1, Yue Yu1, Mark I-Cheng Chen2,3*,
Ee Hui Goh2, Lili Jiang2, Joash Wen Chen Chong2, Yee Sin Leo1,2,3,4, Tau Hong Lee3, Chia Siong Wong3,
Victor Weng Keong Loh5, Adrian Zhongxian Poh6, Tat Yean Tham1,5,6, Wei Mon Wong5,7,8 and Fong Seng Lim1,5

Abstract
Background: Patients’ expectations can influence antibiotic prescription by primary healthcare physicians. We assessed
knowledge, attitude and practices towards antibiotic use for upper respiratory tract infections (URTIs), and whether
knowledge is associated with increased expectations for antibiotics among patients visiting primary healthcare services
in Singapore.
Methods: Data was collected through a cross-sectional interviewer-assisted survey of patients aged ≥21 years waiting
to see primary healthcare practitioners for one or more symptoms suggestive of URTI (cough, sore throat, runny nose
or blocked nose) for 7 days or less, covering the demographics, presenting symptoms, knowledge, attitudes, beliefs
and practices of URTI and associated antibiotic use. Univariate and multivariate logistic regression was used to assess
independent factors associated with patients’ expectations for antibiotics.
Results: Nine hundred fourteen out of 987 eligible patients consulting 35 doctors were recruited from 24 private sector
primary care clinics in Singapore. A third (307/907) expected antibiotics, of which a substantial proportion would ask the
doctor for antibiotics (121/304, 40 %) and/or see another doctor (31/304, 10 %) if antibiotics were not prescribed. The
majority agreed “antibiotics are effective against viruses” (715/914, 78 %) and that “antibiotics cure URTI faster” (594/912,
65 %). Inappropriate antibiotic practices include “keeping antibiotics stock at home” (125/913, 12 %), “taking leftover
antibiotics” (114/913, 14 %) and giving antibiotics to family members (62/913, 7 %). On multivariate regression, the
following factors were independently associated with wanting antibiotics (odds ratio; 95 % confidence interval): Malay
ethnicity (1.67; 1.00–2.79), living in private housing (1.69; 1.13–2.51), presence of sore throat (1.50; 1.07–2.10) or fever (1.46;
1.01–2.12), perception that illness is serious (1.70; 1.27–2.27), belief that antibiotics cure URTI faster (5.35; 3.76–7.62) and not
knowing URTI resolves on its own (2.18; 1.08–2.06), while post-secondary education (0.67; 0.48–0.94) was inversely
associated. Those with lower educational levels were significantly more likely to have multiple misconceptions about
antibiotics.
(Continued on next page)

* Correspondence: mark.chen.ic@gmail.com
†
Equal contributors
2
Saw Swee Hock School of Public Health, National University Health System,
National University of Singapore, 12 Science Drive 2, 117549 Singapore,
Singapore
3
Institute of Infectious Diseases & Epidemiology, Communicable Disease
Centre, Tan Tock Seng Hospital, 308433 Singapore, Singapore
Full list of author information is available at the end of the article

© The Author(s). 2016 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Pan et al. BMC Family Practice (2016) 17:148 Page 2 of 9

(Continued from previous page)

Conclusion: Majority of patients seeking primary health care in Singapore are misinformed about the role of antibiotics in
URTI. Agreeing with the statement that antibiotics cure URTI faster was most strongly associated with wanting antibiotics.
Those with higher educational levels were less likely to want antibiotics, while those with lower educational levels more
likely to have incorrect knowledge.
Keywords: Antibiotic use, Upper Respiratory Tract Infections (URTIs), Primary healthcare, Singapore, Educational level

Background identify possible strategies and specific areas for health

The rise in antibiotic resistance has become an increasing education on better antibiotic stewardship. Specifically,
public health concern worldwide [1]. In Singapore, patients our study aimed to describe the prevalence of miscon-
admitted to local public hospitals have one of the highest ceptions about URTIs and antibiotics, and identify key
rates of antimicrobial resistance worldwide [2, 3], with local misconceptions associated with inappropriately wanting
use of oral antibiotics in the community shown to be asso- antibiotics. We also assessed if any misconceptions were
ciated with increased colonization with Extended-Spectrum especially prevalent in particular population subgroups.
Beta-Lactamase (ESBL) Gram-Negative Bacteria on admis-
sion [4]. The impact of antibiotic resistance include Methods
increased morbidity and mortality from antibiotic-resistant Study design and setting
infections [5], increased socioeconomic burden and greater We conducted a cross-sectional study over eight work-
healthcare costs [1, 6]. ing days in February 2015. Patients were recruited from
Poor antibiotic stewardship is a key driver of antibiotic those seeing 35 GPs at 24 clinics of various sizes, includ-
resistance [7]. A substantial proportion of all antibiotics ing both solo and group practices across Singapore in
are prescribed in the community [8], and Upper Respira- both residential and commercial areas. We were referred
tory Tract Infection (URTI) is one of the commonest to these GPs mainly through the academic medicine net-
conditions in the primary care setting for which anti- work affiliated with the National University of Singapore.
biotic prescriptions have been reported to be high Following initial contact via emails or phone calls, site
worldwide [9–11]. In Singapore, URTIs account for a visits were conducted for interested GPs where the study
quarter of all primary care attendances [12]. While there objectives and execution were explained. Consent was
is no local data on antibiotic use for URTIs in Singapore, obtained to survey patients at each clinic.
it has been noted that antibiotics are commonly pre- Fieldwork was conducted by a team of 38 fourth-year
scribed for URTIs [13]. However, current evidence-based medical students from the National University of
guidelines do not support antibiotic use in the majority Singapore (NUS) Yong Loo Lin School of Medicine (YLL-
of URTI cases [14, 15], as URTIs are frequently of viral SoM) who were deployed in pairs to participating clinics
etiology [16–18]. are often self-limiting [19, 20], and during operating hours. These students underwent a care-
seldom lead to serious complications [21]. fully planned full day training program including video
Inappropriate expectations of antibiotics by patients demonstrations, simulation and role-play to familiarize
have been commonly observed in primary healthcare, them with the study protocol and standardize the process
and is a key factor driving over-prescription of antibi- of administering the questionnaire.
otics in such settings. For instance, Linder et al found We aimed to study all patients aged 21 years and
that physicians are more likely to prescribe antibiotics to above, presenting with at least one of four URTI symp-
patients who desire antibiotics [22]. Furthermore, Scott toms (runny nose, blocked nose, cough or sore throat)
et al observed that various inappropriate behaviours by for seven days or less at participating clinics. Patients
patients often pressured physicians to prescribe antibi- were excluded if they had sought medical consultation
otics [23], such as direct request for antibiotics, portray- for the same symptoms in the preceding 30 days, were
ing severity of illness, or volunteering previous positive on long-term immunosuppressive or oral corticosteroid
experience with use of antibiotics. Lam et al also ob- medications, had chronic kidney disease, had a past
served that primary healthcare physicians over-prescribe history of advanced stage or metastatic cancer, were
antibiotics in order to satisfy their patients [24]. These immunocompromised (e.g. human immunodeficiency
studies underscore how patient’s expectations for antibi- virus infection), or were not conversant in English or
otics influence prescriptions by physicians. Mandarin. Eligible patients who provided written con-
In Singapore, the majority of URTI patients (87 %) are sent were enrolled into the study.
seen in private sector general practitioner (GP) clinics Following enrolment, researchers administered an
[12]. We hence surveyed patients of GP clinics to interviewer-assisted pre-consultation questionnaire, in
Pan et al. BMC Family Practice (2016) 17:148 Page 3 of 9

the clinics’ waiting rooms. The questionnaire was replied as “not sure” or gave the incorrect answer were
designed to include several factors identified from other grouped as having an incorrect response.
studies, in particular a previous study done in Singapore We also investigated factors associated with whether a
[13]. During the design process, we engaged in consulta- patient wanted antibiotics as a key outcome of interest,
tions with a panel of experts including primary care phy- based on the response to the question “I want to receive
sicians, infectious disease experts and public health antibiotics”. As above, those who responded as “strongly
experts to guide questionnaire development. The draft agree” and “agree” were considered as wanting antibi-
questionnaire was then piloted in a group of lay person otics and vice-versa for “strongly disagree” and “dis-
volunteers before being field tested in a group of five agree”. Logistic regression, with robust standard errors
clinics with actual patients to assess their understanding to account for potential clustering of results at the GP
of individual questionnaire items. Inputs from these pa- level, was performed to ascertain factors associated with
tients and their physicians were then used to refine the wanting antibiotics, including participant’s sociodemo-
phrasing of the questionnaire. The questionnaire elicited graphic factors (age, gender, ethnicity, employment,
details about the patients’ demographics, current episode housing), episodic factors (symptom duration, presenting
of illness, knowledge, attitudes and beliefs about URTI symptoms, payment mode), perception of illness sever-
and antibiotics, antibiotic practices and health-seeking ity, and questions pertaining to knowledge and beliefs
behavior including wanting antibiotics. In order to elicit about antibiotic use. Univariate and multivariate odds
responses about URTI from lay participants, we referred ratios (ORs) with 95 % confidence intervals (CIs) were
to URTI as respiratory infection with common cough calculated, with the multivariate estimates adjusting for
and flu symptoms in the questionnaire. all other covariates that were also assessed in univariate
analysis. The exception was our decision to exclude re-
Power calculations sponses reflecting practices (e.g. “I take leftover antibi-
Based on the available manpower resources and study otics when I have similar symptoms”), as these were
timeframe, we projected recruitment of up to 1000 par- statements describing behaviours which might arise
ticipants for this study. Assuming that approximately from the same underlying motivations which make pa-
40 % of patients would want antibiotics based on exist- tients want antibiotics. However, we assessed the associ-
ing literature [22, 25], this gave an estimated margin of ation between these inappropriate practices as well as
error of 3 % at 95 % confidence level in estimating the with wanting antibiotics, separately presenting phi coef-
proportion of patients who expected antibiotics. It would ficients from Pearson’s correlation as a measure of asso-
also give us a power of 92 % to detect, at p < 0.05, fac- ciation between these factors. Finally, we also assessed if
tors that are at least 20 % more common in those who particular beliefs or incorrect knowledge might be espe-
expected antibiotics as compared to those who did not. cially prevalent in particular sociodemographic sub-
groups, presenting p-values from chi-squared and
Data management and analysis Fisher’s exact tests.
Data collected at each GP clinic was double-entered into All data was analyzed using Stata for Windows, ver-
a shared database. Frequency tabulations were per- sion 11 (Stata Corporation, College Station, Texas,
formed for all descriptive data, with 95 % confidence USA), with p-values of less than 0.05 considered statisti-
intervals presented where relevant. cally significant.
To facilitate interpretation, we dichotomised the re-
sponses to several survey items. For questions on atti- Ethics approval
tudes, beliefs and practices, participants’s agreement to a The study was approved by the institutional review
given statement was measured on a 4-point scale from board of the National University of Singapore (reference
“strongly disagree” to “strongly agree” (e.g. “I believe that B-14-259).
antibiotics cure my respiratory infection faster”). The
participants who responded as “strongly agree” and Results
“agree” were grouped as agreeing, while those who Out of a total of 987 eligible patients, 914 patients gave
“strongly disagree” and “disagree” were considered as signed informed consent to participate in the study (re-
disagreeing with the statement. In survey items assessing sponse rate = 92.6 %). Table 1 compares their socio-
knowledge, participants could either answer “yes”, “no” demographic profiles against 2014 population trends for
or “not sure” in response to a given statement where Singapore [26]. The median age of participants was
there was a designated correct answer (e.g. “Viruses 35 years, and there were approximately equal numbers
cause most respiratory infections” where the correct an- of males and females. Our study had similar proportions
swer is yes). Those who answered correctly were consid- of each major ethnic group compared to that of the gen-
ered as giving the appropriate response, while those who eral Singapore residential population, though there were
Pan et al. BMC Family Practice (2016) 17:148 Page 4 of 9

Table 1 Characteristics of study participants (N = 914) in comparison with 2014 Population Trends
Characteristic Study participants (N = 914) Singapore residentsa,d
No. (%) (N = 3,870,739)
Age in years Median 35.0 39.3
21–34 438 (47.9) 821,864 (21.2)
35–49 308 (33.7) 926,585 (23.9)
50–64 121 (13.2) 835,397 (21.6)
≥65 47 (5.1) 431,601 (11.2)
Gender Male 443 (48.5) 1,900,513 (49.1)
Female 471 (51.5) 1,970,186 (50.9)
Ethnicity Chinese 630 (68.9) 2,874,380 (74.3)
Malay 116 (12.7) 516,657 (13.3)
Indian 99 (10.8) 353,021 (9.1)
Other 69 (7.6) 126,681 (3.3)
b c
Highest qualification attained Primary and below 47 (5.1) 833,300 (31.2)
Secondary 187 (20.5) 501,200 (18.8)
Post-secondary 679 (74.4) 1334,700 (50.0)
Employment status Currently employed 781 (85.4) -
Not currently employed 70 (7.7) -
Student 63 (6.9) -
Housing typeb Public Housing 745 (81.7) 3154,691 (81.5)
Private Housing 167 (18.3) 678,808 (17.5)
Mode of payment Full Payment 275 (30.1) -
Partial Subsidy 385 (42.1) -
Full Subsidy 254 (27.8) -
a
Singaporeans and Singapore permanent residents only. Data taken from Singapore population trends 2014 unless specified
b
Excludes 1 observation with missing data on highest qualification attained and 2 observations with missing data on housing type
c
Equivalent to 6 years of formal education or less
d
Department of Statistics Singapore. Population Trends 2014. Singapore: 2014

slightly less Chinese and more participants of other eth- majority (79.5 %, 727/914) did at least know using anti-
nic groups in our study. Compared to the general popu- biotics can result in lack of effectiveness in the long
lation, study participants were of a higher education term. Close to two thirds (65.1 %, 594/912) agreed that
level with a lower proportion having primary education antibiotics cure URTI faster, while a third (33.8 %, 307/
and below, and a higher proportion with post-secondary 907) wanted antibiotics (Fig. 1c), of which 39.8 % (121/
education. The proportions staying in public and private 304) would ask the doctor for antibiotics if not given;
housing were similar to that of the general population. however, only 8.6 % (26/304) would not accept the doc-
Majority of our patients had partial or full subsidy of the tor’s decision if the doctor explains why antibiotics were
payment for that visit from either pre-paid insurance or not prescribed, and 10.2 % (31/304) would see another
government subsidy schemes. doctor (Fig. 1d). Figure 1c also shows the prevalence of
Our patients presented to the clinics with mainly poor antibiotic practices, which include: “keeping anti-
symptoms of cough, sore throat and runny or blocked biotic stock at home” (125/913, 13.7 %), “taking leftover
nose, with duration of illness mostly between 1 to 4 days; antibiotics” (114/913, 12.4 %) and giving antibiotics to
18.2 % of patients were worried that their illness was family members (62/913, 6.8 %). These practices were
something serious (Fig. 1a). 32.6 % (298/913) did not significantly correlated with each other at p < 0.001 (phi
know that viruses are the cause of most URTIs, but coefficients 0.40–0.50) and with wanting antibiotics
nearly half (48.8 %, 446/914) did not know URTI re- (Table 2).
solves on its own, and 78.3 % (715/914) did not reject Figure 2 shows factors associated with wanting antibiotics
the statement that antibiotics were effective against vi- (ORs, 95 % CIs). Indians (1.77, 1.15–2.74) and Malays
ruses (Fig. 1b). Also, nearly half (44.0 %, 402/914) did (2.13, 1.42–3.18) were significantly more likely to want anti-
not know antibiotics have side effects, though the biotics than Chinese individuals in the univariate analysis,
Pan et al. BMC Family Practice (2016) 17:148 Page 5 of 9

A. Symptoms and perception of illness severity B. Knowledge about URTI and antibiotic use
Viruses cause most respiratory infections
Symptom duration
(N=913)
- <1 to 2 days (N=914)
Respiratory infection resolves on its own
- 3 to 4 days (N=914) (N=914)
- 5 to 7 days (N=914) †Antibiotics are effective against viruses
Symptoms experienced (N=914)
- cough (N=914) Bacteria that normally live on the skin and in
- sore throat (N=914) the gut are good for health (N=914)
- runny / blocked nose (N=914) Infection by antibiotic resistant bacteria cannot †Questions where
- fever (N=914) be easily cured or cannot be cured (N=914) "No" is considered
the correct answer; in
Worried illness is serious (N=914) †Antibiotics do not cause side effects e.g. all others, "Yes" was
diarrhea, vomiting, allergic reaction (N=914) considered the
0% 20% 40% 60% 80% correct answer.
If I use too much antibiotics, it can result in
% of participants answering yes / them losing effectiveness in long term (N=914)
% with that duration of symptoms
0% 20% 40% 60% 80% 100%
% of participants with inappropriate response

C. Attitudes, beliefs and practices pertaining to antibiotics D. Self-reported reaction if antibiotics are not prescribed
(based on those who wanted antibiotics)
I believe that antibiotics cure my
respiratory infection faster (N=912) Will ask doctor for antibiotics if not offered
(N=304)
I want to receive antibiotics (N=907) ‡Original phrasing was "I will accept
the doctor’s decision to not
I take leftover antibiotics when I Would not acccept doctor's decision‡ (N=304) prescribe medicine if he explains
have similar symptoms (N=913) why"; the question phrasing and
I normally keep antibiotic stock at Will see another doctor if antibiotics are not responses are inverted to facilitate
prescribed (N=304) presentation
home in case of emergency (N=913)
If my family is sick, I usually give my 0% 20% 40% 60%
antibiotic to them (N=913)
% of participants who agree
0% 20% 40% 60% 80%
% of participants who agree

Fig. 1 Prevalence of symptoms, knowledge, attitudes and practices among study population, which includes prevalence of a Type of symptom
presentations and perception of illness severity, b Knowledge about URTI and antibiotic use, c Attitudes, beliefs and practices pertaining to
antibiotics and d Self-reported response among patients who wanted antibiotics if antibiotics are not prescribed to them

but the association remained significant only for Malays viruses (1.91, 1.33–2.74) and that “bacteria that normally
after adjusting for covariates (1.67, 1.00–2.79, Fig. 2a). live on the skin and in the gut are good for health” (1.75,
Multivariate analysis also suggests that older age groups (vs 1.22–2.51) were significant on univariate but not multivari-
those aged 21–34, borderline significant) and participants ate analysis (Fig. 2c). However, patients who were worried
with post-secondary education (0.67, 0.48–0.94, p = 0.021 their illness was serious (1.70, 1.27–2.27), believed that anti-
vs secondary education as reference category) were signifi- biotics cure URTI faster (5.35, 3.76–7.62) and did not know
cantly less likely, while those who lived in private housing that URTI resolves on its own (2.18, 1.08–2.06) or that
(vs public housing) were significantly more likely (1.69, using too much antibiotics resulted in their ineffectiveness
1.13–2.51) to want antibiotics. From Fig. 2b, patients having over the long-term (1.47, 1.00–2.14) were still significantly
symptoms lasting 3–4 days (vs 1–2 days, 0.52, 0.34–0.78) more likely to want antibiotics on multivariate analysis.
were significantly less likely to want antibiotics, while Other factors analyzed in Fig. 2 were not significantly asso-
patients having symptoms of sore throat (1.50, 1.07–2.10) ciated with wanting antibiotics on neither univariate nor
or fever (1.46, 1.01–2.12) were significantly more likely to multivariate analysis.
want antibiotics on multivariate analysis. Multiple miscon- Table 3 presents a stratified analysis by age, gender, ethni-
ceptions were significantly associated with wanting antibi- city and education for the belief that antibiotics cure
otics. Not knowing that antibiotics are not effective against respiratory infection faster, as well as knowledge questions

Table 2 Correlation between inappropriate practices and wanting to receive antibiotics for URTI
Inappropriate practices and wanting phi-coefficient (p-value)
antibiotics
I take leftover antibiotics when I I normally keep antibiotic stock at If my family is sick, I usually give
have similar symptoms home in case of emergency my antibiotic to them
I normally keep antibiotic stock at 0.495 (<0.001) - -
home in case of emergency
If my family is sick, I usually give my 0.425 (<0.001) 0.399 (<0.001) -
antibiotic to them
I want to receive antibiotics 0.172 (<0.001) 0.175 (<0.001) 0.130 (<0.001)
Pan et al. BMC Family Practice (2016) 17:148 Page 6 of 9

A. Sociodemographic factors B. Mode of payment and symptoms

Age in years (vs 21-34) Mode of payment (vs full payment)
35-49 Partial Subsidy
50-64
≥65 Full Subsidy

Female (vs male) Symptom duration (vs 1 to 2 days)

3 to 4 days
Ethnicity (vs Chinese) 5 to 7 days
Malay
Indian
Other Has symptom (vs none)
Cough
Highest qualification (vs sec.) Sorethroat
Pri. and below Runny / Blocked nose
Post-sec.
Fever
Employment (vs currently employed)
Not currently employed 0.1 1 10
Odds ratio
Student
Private (vs public housing) Legend
0.1 1 10 Univariate
Odds ratio Multivariate

C. Perception of illness severity and beliefs and knowledge about antibiotic use

Worried that illness is something serious (vs no)

I believe that antibiotics cure my respiratory infection faster (vs no)
Viruses cause most respiratory infections†
Respiratory infection resolves on its own†
Antibiotics are effective against viruses†
Bacteria that normally live on the skin and in the gut are good for health†
Infection by antibiotic resistant bacteria cannot be easily cured or cannot be cured†
Antibiotics do not cause side effects e.g. diarrhea, vomiting, allergic reactions†
If I use too much antibiotics, it can result in them losing effectiveness in long term†

†reference category is those with appropriate response 0.1 1 10

Odds ratio
Fig. 2 Factors associated with wanting antibiotics on univariate and multivariate analysis. a Sociodemographic factors; b Mode of payment and
symptoms; c Perception of illness severity and beliefs and knowledge about antibiotic use. Diamonds give the odds ratio on univariate analysis
(blue) and multivariate analysis (red), with 95 % intervals as error bars

significantly associated on either univariate or multivariate believed that antibiotics cure URTI faster, and 74.5 % did
analysis with wanting antibiotics. There was a significant not reject the statement that antibiotics are effective against
difference among age groups for the questions “Respiratory viruses.
infection resolves on its own” (p = 0.042), and “Bacteria that
normally live on the skin and in the gut are good for health” Discussion
(p <0.001), with those ≥65 years old being more likely to Majority of our study patients seeking primary health
respond incorrectly than other age groups. However, this care in Singapore are misinformed about the role of
effect was found on bivariate analysis to be related to edu- antibiotics in URTI. In particular, poor knowledge was
cational level. There were also statistically significant but highly prevalent and associated with wanting antibiotics.
modest differences in proportion of males versus females A third wanted antibiotics, and more than half believed
who responded incorrectly to the above two questions, and antibiotics could cure their illness faster and had serious
Malays were significantly more likely not to know that “an- misconceptions with regards to the effectiveness of anti-
tibiotics are effective against viruses” (p = 0.002); these biotics for URTI. A substantial minority also had poor
effects persisted when accounting for educational levels. and potentially dangerous antibiotic practices such as
Patients with lower educational levels, especially the group using left-over antibiotics for themselves and family
with primary education and below, were significantly more members. Agreeing that antibiotics cure URTI faster was
likely to believe that antibiotics cure respiratory infection most strongly associated with wanting antibiotics, and
faster, as well as have incorrect responses to all four know- educational level was significantly associated with this
ledge questions analysed. However, even amongst those belief as well as incorrect knowledge about antibiotics
with higher education, 63.2 % of the post-secondary group and URTI.
Pan et al. BMC Family Practice (2016) 17:148 Page 7 of 9

Table 3 Sociodemographic variables associated with believing that antibiotics cure respiratory infection faster and incorrect
responses to key knowledge questions
Sociodemographic strata Believe that Incorrect response to:
antibiotics cure
“Respiratory “Antibiotics are “Bacteria that “If I use too much
respiratory
infection resolves effective against normally live on antibiotics, it can
infection faster
on its own” viruses” the skin and in the result in them losing
gut are good for effectiveness in the
health” long-term”
No. (%) p-value No. (%) p-value No. (%) p-value No. (%) p-value No. (%) p-value
Age in years 21–34 288 (65.9) 0.804† 203 (46.3) 0.042† 348 (79.5) 0.493† 58 (13.2) <0.001† 91 (20.8) 0.318†
35–49 200 (64.9) 150 (48.7) 226 (73.4) 47 (15.3) 55 (17.9)
50–64 73 (60.3) 65 (53.7) 101 (83.5) 18 (14.9) 26 (21.5)
≥65 33 (71.7) 28 (59.6) 40 (85.1) 22 (46.8) 15 (31.9)
Gender Male 285 (64.5) 0.728* 200 (45.1) 0.034* 359 (81.0) 0.054* 84 (19.0) 0.014* 96 (21.7) 0.412*
Female 309 (65.7) 246 (52.2) 356 (75.6) 61 (13.0) 91 (19.3)
Ethnicity Chinese 388 (61.6) 0.006* 297 (47.1) 0.157* 486 (77.1) 0.002* 94 (14.9) 0.065* 118 (18.7) 0.073*
Malay 89 (76.7) 65 (56.0) 105 (90.5) 24 (20.7) 33 (28.4)
Indian 69 (71.1) 54 (54.5) 77 (77.8) 21 (21.2) 24 (24.2)
Other 48 (69.6) 30 (43.5) 47 (68.1) 6 (8.7) 12 (17.4)
Highest qualification attained Pri. or less 35 (74.5) 0.028† 28 (59.6) <0.001† 45 (95.7) <0.001 24 (51.1) <0.001† 20 (42.6) <0.001†
Sec. 131 (70.1) 112 (59.9) 163 (87.2) 55 (29.4) 62 (33.2)
Post-sec. 428 (63.2) 305 (44.9) 506 (74.5) 66 (9.7) 105 (15.5)
† p-value for Chi-Square test for trend
* p-value for Chi-Square test

The proportion of our study participants who wanted Patients who were worried their illness was something ser-
antibiotics was comparable to those of other studies in ious were also 1.7 times more likely to want antibiotics,
Hong Kong and Boston [22, 25], where the proportion which is comparable to a study in Minnesota [29] where
who wanted antibiotics ranged from 36 to 39 % respect- those who rated their symptoms as severe were twice as
ively, and the proportion with poor antibiotic practices likely to want antibiotics. Doctors could help address
was also comparable to those from a similar study in perceptions that fever and sore throat warrant antibiotics,
Malaysia [27]. However, the prevalence of misconcep- since these could be commonly due to viral infections, as
tions in our study was substantially higher than in a well as reassure worried patients about the benign nature of
similar study in Minnesota [28]; For example, half our their illness where appropriate to alleviate unnecessary
study participants did not know that URTI resolves on fears and thus reduce the want for antibiotics.
its own, as compared to 15 % in that study. On the other Regarding sociodemographic factors, younger patients,
hand, the prevalence of misconceptions was comparable Malays, and participants living in private housing were
with a previous local study in Singapore by Tan et al among more likely to want antibiotics. The reasons are unclear,
URTI patients seeking medical care in the public primary but may be due to residual confounding from inappro-
care clinics [13]. For example, the belief that antibiotics priate beliefs or attitudes that we did not measure or ad-
cure URTI faster was similarly common (61.7 % [Tan et al] equately adjust for. Those with higher educational levels
vs 65.1 % [our study]). This belief was also most strongly were less likely to want antibiotics, similar to findings
associated with wanting antibiotics, while knowing that vi- from a study from Boston [22]. This effect persisted after
ruses caused most respiratory infections was not associated adjusting for differences in prevalence of misconceptions
with wanting antibiotics. On the other hand, patients who between educational levels. We did also find a strong as-
knew that URTI resolves on its own were significantly less sociation between lower educational levels and higher
likely to want antibiotics. The content of health education probability of believing antibiotics cure their respiratory
messaging must hence go beyond educating about the infection faster and having incorrect knowledge on anti-
causes of URTI to emphasizing its self-limiting nature and biotics, and this finding is consistent with other studies
the ineffectiveness of antibiotics against viruses. in Malaysia, Korea and Hong Kong [27, 30–32]. Educa-
With regards to presenting symptoms and self-perceived tional level may act through the specific items of know-
illness severity, we found that patients having sore throat or ledge and beliefs covered in our study, as well as other
fever were significantly more likely to want antibiotics. pathways not measured here. Notably, knowledge about
Pan et al. BMC Family Practice (2016) 17:148 Page 8 of 9

antibiotics is not commonly taught during foundation promoting better antibiotic stewardship, although
primary schooling years. Less educated participants may further research is needed to ascertain what type of
also have less access to health education through other interventions would be effective.
information channels, or face language barriers in under-
standing health education materials. However, given that Additional files
incorrect knowledge and the belief that antibiotics cure
respiratory infections faster is highly prevalent regardless Additional file 1: Preconsultation Questionnaire for Patients. The
pre-consultation questionnaire for patients. Contains a Chinese translation.
of educational levels, and the link established between (DOCX 50 kb)
these factors and wanting antibiotics, we believe that Additional file 2: Dataset Used. The dataset collected through the
better public education is important in the fight against surveys. Includes the dataset used in the analysis. (XLSX 317 kb)
inappropriate antibiotic prescriptions in Singapore. Additional file 3: Codebook for dataset supplied. The codebook for
Interventions could take the form of patient education interpretation of Material B, the data provided. (XLSX 11 kb)
through videos and pamphlets or individualized counsel-
ing at primary care clinics, although the efficacy and feasi- Abbreviations
95 % CI: 95 % Confidence interval; GPs: General practitioners; NUS: National
bility of such interventions in Singapore has not been University of Singapore; OR: Odds ratio; URTIs: Upper respiratory throat
studied and needs to be investigated. Alternatives to infections; YLLSoM: Yong Loo Lin School of Medicine
clinic-based education would be community-wide public
education targeting the entire society; for instance, school- Acknowledgement
We would like to thank all participating GPs for offering their clinics for
based health education programs have been shown to sig- patient recruitment. We also acknowledge other members of the Yong Loo
nificantly increase antibiotic-related knowledge among Lin School of Medicine 2014/2015 Community Health Project Team 1 for
middle-school children in Portugal [33] and Moldova [34]. their contributions to study design and conduct of field work, and Dr Gerald
Koh and Ms Moira Soh, for providing academic and administrative support
There are currently no structured health education pro- during the project. We also thank Ms Hsu Jung Pu for providing editorial
grams on antibiotic use in the public primary or secondary support for the manuscript. Finally, we acknowledge the Saw Swee Hock
school syllabus in Singapore. However, the “awareness that School of Public Health for funding the study.

bacteria can have both beneficial and harmful effects” has Funding
been part of the common lower secondary science Not applicable.
syllabus in Singapore [35]; This may be why only 10 % of
Availability of data and materials
those who had post-secondary education gave an incorrect The dataset, the corresponding codebook for interpretation and the
response to our question about normal bacterial flora, questionnaire supporting the conclusions of this article are included within
compared to more than half of those with primary educa- the article (and its Additional files 1, 2 and 3).
tion or less. Therefore, a health education program inte- Authors’ contributions
grated into the school curriculum which promotes DSP and JHH participated in the data collection and led the writing of the
rational use of antibiotics in the community from young manuscript writing. MHL and YY participated in the data collection and data
analysis. MIC, EHG, LJ, and YSL participated in in study design, statistical
may be beneficial. Future research could hence look into analysis and manuscript writing. JWC participated in data analysis,
the effectiveness of both clinic and community-wide manuscript editing and writing. THL, CSW, VWL, AZP, TYT, WMW and FSL
health education programs on proper use of antibiotics. participated in patient recruitment and manuscript writing. All authors read
and approved the final manuscript.
One limitation we acknowledge was selection bias in
our recruitment of GPs. Due to resource constraints, the Competing interests
number of GP clinics in our study was relatively small The authors declare that they have no competing interests.
compared to the total number of GP clinics in Singapore
Consent for publication
(24 vs 1268 [36]), and recruited from our existing contacts Not applicable.
including an academic family medicine network. While it
would have been ideal to recruit a more representative Ethics approval and consent to participate
Ethics approval was obtained via the National University of Singapore Institutional
sample of GP clinics, another study previously attempted Review Board (NUS-IRB) with approval reference number: B-14-259. All participants
this and encountered dismal response rates [37]. in the study had consented and participated in the study of their own volition.

Author details
Conclusion 1
Yong Loo Lin School of Medicine, National University Health System,
Our results suggest a substantial proportion of National University of Singapore, 119228 Singapore, Singapore. 2Saw Swee
patients in Singapore are misinformed on the role of Hock School of Public Health, National University Health System, National
University of Singapore, 12 Science Drive 2, 117549 Singapore, Singapore.
antibiotics in URTI. Incorrect knowledge about antibi- 3
Institute of Infectious Diseases & Epidemiology, Communicable Disease
otics, and the belief that antibiotics cure URTI faster Centre, Tan Tock Seng Hospital, 308433 Singapore, Singapore. 4Lee Kong
was highly prevalent, with the latter being strongly Chian School of Medicine, Nan Yang Technological University, 308232
Singapore, Singapore. 5Division of Family Medicine, Department of Medicine,
associated with wanting antibiotics. Possible interven- University Medicine Cluster, National University Hospital System, 119228
tions include clinic and community-based education Singapore, Singapore. 6Frontier Healthcare group, 400305 Singapore,
Pan et al. BMC Family Practice (2016) 17:148 Page 9 of 9

Singapore. 7Division of Primary Care, Raffles Medical Group, 188770 22. Linder JA, Singer DE. Desire for antibiotics and antibiotic prescribing for
Singapore, Singapore. 8Duke NUS Graduate Medical School, National adults with upper respiratory tract infections. J Gen Intern Med. 2003;
University of Singapore, 169857 Singapore, Singapore. 18(10):795–801.
23. Scott JG, Cohen D, Dicicco-Bloom B, Orzano AJ, Jaen CR, Crabtree BF.
Received: 15 October 2015 Accepted: 21 October 2016 Antibiotic use in acute respiratory infections and the ways patients pressure
physicians for a prescription. J Fam Pract. 2001;50(10):853–8.
24. Lam T, Lam K. What are the non‐biomedical reasons which make family
doctors over‐prescribe antibiotics for upper respiratory tract infection in a
References mixed private/public Asian setting? J Clin Pharm Ther. 2003;28(3):197–201.
1. World Health Organization. Antimicrobial resistance: global report on 25. Chan CS. What do patients expect from consultations for upper respiratory
surveillance. France: World Health Organization; 2014. tract infections? Fam Pract. 1996;13(3):229–35.
2. Hsu LY, Kwa AL, Lye DC, Chlebicki MP, Tan TY, Ling ML, Wong SY, Goh LG. 26. Department of Statistics S. Population Trends 2014. Singapore: Department
Reducing antimicrobial resistance through appropriate antibiotic usage in of Statistics Singapore; 2014.
Singapore. Singapore Med J. 2008;49(10):749–55. 27. Lim KK, Teh CC. A Cross sectional study of public knowledge and attitude
3. Hsu LY, Tan TY, Jureen R, Koh TH, Krishnan P, Tzer-Pin Lin R, Wen-Sin Tee N, towards antibiotics in Putrajaya, Malaysia. Southern Med Rev. 2012;5(2):26.
Tambyah PA. Antimicrobial drug resistance in Singapore hospitals. Emerg 28. Braun BL, Fowles JB, Solberg L, Kind E, Healey M, Anderson R. Patient beliefs
Infect Dis. 2007;13(12):1944–7. about the characteristics, causes, and care of the common cold. J Fam
4. Pada S, Lye D, Krishnan P, Chan J, Chan S, Cham G, Ang B, Leo Y. Prevalence Pract. 2000;49(2):153–6.
and predictors of methicillin-resistant Staphylococcus aureus (MRSA) and 29. Braun BL, Fowles JB. Characteristics and experiences of parents and adults
extended-spectrum beta-lactamase (ESBL) Gram-negative bacteria at hospital who want antibiotics for cold symptoms. Arch Fam Med. 2000;9(7):589.
presentation in Singapore. Int J Infect Dis. 2008;12:e449–50. 30. Kim SS, Moon S, Kim EJ. Public knowledge and attitudes regarding
5. Butler CC, Hillier S, Roberts Z, Dunstan F, Howard A, Palmer S. Antibiotic-resistant antibiotic use in South Korea. J Korean Acad Nurs. 2011;41(6):742–9.
infections in primary care are symptomatic for longer and increase workload: 31. Oh AL, Hassali MA, Al-Haddad MS, Sulaiman SAS, Shafie AA, Awaisu A.
outcomes for patients with E. coli UTIs. Br J Gen Pract. 2006;56(530):686–92. Public knowledge and attitudes towards antibiotic usage: a cross-sectional
6. Thabit AK, Crandon JL, Nicolau DP. Antimicrobial resistance: impact on study among the general public in the state of Penang, Malaysia. J Infect
clinical and economic outcomes and the need for new antimicrobials. Dev Ctries. 2010;5(05):338–47.
Expert Opin Pharmacother. 2015;16(2):159–77. 32. You J, Yau B, Choi K, Chau C, Huang Q, Lee S. Public knowledge, attitudes
7. Soothill G, Hu Y, Coates A. Can we prevent antimicrobial resistance by using and behavior on antibiotic use: a telephone survey in Hong Kong. Infection.
antimicrobials better? Pathogens. 2013;2(2):422–35. 2008;36(2):153–7.
8. Wise R, Hart T, Cars O, Streulens M, Helmuth R, Huovinen P, Sprenger M. 33. Azevedo M-M, Pinheiro C, Yaphe J, Baltazar F. Assessing the impact of a
Antimicrobial resistance: is a major threat to public health. BMJ (Br Med J). school intervention to promote students’ knowledge and practices on
1998;317(7159):609. correct antibiotic use. Int J Environ Res Public Health. 2013;10(7):2920–31.
9. Goossens H, Ferech M, Vander Stichele R, Elseviers M, Group EP. Outpatient 34. Cebotarenco N, Bush PJ. Reducing antibiotics for colds and flu: a
antibiotic use in Europe and association with resistance: a cross-national student-taught program. Health Educ Res. 2008;23(1):146–57.
database study. Lancet. 2005;365(9459):579–87. 35. Curriculum Planning & Development Division. Lower Secondary Science
10. McCaig LF, Hughes JM. Trends in antimicrobial drug prescribing among Syllabus for Express/Normal (Academic). Singapore: Ministry of Education
office-based physicians in the United States. JAMA. 1995;273(3):214–9. Singapore; 2012.
11. Trap B, Hansen E. Treatment of upper respiratory tract infections–a 36. Wong CY TK, Foo G, Chua A, Lee YV, Wong TH, Chong YW. 2006 survey of
comparative study of dispensing and non‐dispensing doctors. J Clin Pharm GP clinic practice costs in Singapore. SMA News. 2007;39(11):10–9.
Ther. 2002;27(4):289–98. 37. Ong JBS, Chen M, Cook AR, Lee HC, Lee VJ, Lin RTP, Tambyah PA, Goh LG.
12. Sng QS. Primary Care Survey 2010: Profile of Primary Care Patients. Real-time epidemic monitoring and forecasting of H1N1-2009 using
Singapore: Ministry of Health; 2011. influenza-like illness from general practice and family doctor clinics in
13. Tan Y, Hong C, Chong P, Tan E, Lew Y, Lin R. Knowledge that upper Singapore. PLoS One. 2010;5(4):e10036.
respiratory tract infection resolves on its own is associated with more
appropriate health-seeking behaviour and antibiotic cognition. Singapore
Med J. 2006;47(6):518–24.
14. Zoorob R, Sidani MA, Fremont RD, Kihlberg C. Antibiotic use in acute upper
respiratory tract infections. Am Fam Physician. 2012;86(9):817–22.
15. Centre for Clinical Practice at NICE (UK). Respiratory Tract Infections -
Antibiotic Prescribing: Prescribing of Antibiotics for Self-Limiting Respiratory
Tract Infections in Adults and Children in Primary Care. London: National
Institute for Health and Clinical Excellence (UK); 2008. http://www.ncbi.nlm.
nih.gov/books/NBK53632/. Accessed 24 Oct 2016.
16. Hong CY, Lin RT, Tan ES, Chong PN, Tan YS, Lew YJ, Loo LH. Acute respiratory
symptoms in adults in general practice. Fam Pract. 2004;21(3):317–23.
17. van Gageldonk-Lafeber AB, Heijnen ML, Bartelds AI, Peters MF, van der Plas SM,
Wilbrink B. A case-control study of acute respiratory tract infection in general
practice patients in The Netherlands. Clin Infect Dis. 2005;41(4):490–7.
18. Poole PM, Tobin JO. Viral and epidemiological findings in MRC/PHLS Submit your next manuscript to BioMed Central
surveys of respiratory disease in hospital and general practice. Postgrad and we will help you at every step:
Med J. 1973;49(577):778–87.
19. Gonzales R, Bartlett JG, Besser RE, Cooper RJ, Hickner JM, Hoffman JR, Sande • We accept pre-submission inquiries
MA. Principles of appropriate antibiotic use for treatment of acute • Our selector tool helps you to find the most relevant journal
respiratory tract infections in adults: background, specific aims, and
• We provide round the clock customer support
methods. Ann Emerg Med. 2001;37(6):690–7.
20. Meropol SB, Localio AR, Metlay JP. Risks and benefits associated with • Convenient online submission
antibiotic use for acute respiratory infections: a cohort study. Ann Fam Med. • Thorough peer review
2013;11(2):165–72.
• Inclusion in PubMed and all major indexing services
21. Petersen I, Johnson A, Islam A, Duckworth G, Livermore D, Hayward A.
Protective effect of antibiotics against serious complications of common • Maximum visibility for your research
respiratory tract infections: retrospective cohort study with the UK General
Practice Research Database. BMJ. 2007;335(7627):982. Submit your manuscript at
www.biomedcentral.com/submit