Professional Documents
Culture Documents
Trends of etiology and drug resistance in enteric fever in the last two decades in
cr
ipt
Surendra Karki1, Prabin Shakya2, Allen C Cheng1, Shyam Prakash Dumre3, Karin
Leder1
1
an
Medicine, Monash University, 99 Commercial Road, Melbourne VIC 3004. tel: +61 3
99030118 Email: surendra.karki@monash.edu
pt
ed
Alternate author for contact: Allen C Cheng, Department of Epidemiology and Preventive
Medicine, Monash University. 2nd floor, Burnet Centre, Alfred Hospital, Commercial Road,
Melbourne VIC 3004. tel: +61 3 9076 3009, fax: +61 3 9076 2431, email:
allen.cheng@monash.edu
40 word summary
ce
Over the last two decades in Nepal, there has been a significant trend towards reduced
susceptibility or non-susceptibility to fluoroquinolones. In contrast, decreasing resistance to
Ac
TheAuthor2013.PublishedbyOxfordUniversityPressonbehalfoftheInfectiousDiseasesSocietyof
America.Allrightsreserved.ForPermissions,pleaseemail:journals.permissions@oup.com
us
Abstract
Prospective time-trend analyses on shifting etiology, and trends of drug-resistance in enteric
cr
ipt
fever are scarce. We performed a systematic review and meta-analysis to understand the
trends in etiology, and resistance to antimicrobials since 1993, using published and
unpublished datasets from Nepal. Thirty-two studies involving 21,067 Salmonella enterica
serotype Typhi (ST) and Salmonella enterica serotype Paratyphi A (SPA) isolates were
us
included. There was an increasing trend in enteric fever caused by SPA during the last two
decades (p<0.01). We observed sharply increasing trends in resistance to nalidixic acid, and
an
chloramphenicol, and co-trimoxazole have significantly decreased for both organisms. The
resistance to ceftriaxone has remained low, suggesting it is likely to remain useful as a
reserve antibiotic for treatment. Decreasing resistance to traditional first-line antibiotics and
Ac
ce
pt
ed
ciprofloxacin for both ST and SPA. In contrast, multi-drug resistance (MDR), resistance to
Introduction
The term enteric fever includes Typhoid fever and Paratyphoid fever, caused by
cr
ipt
Salmonella enterica serotype Typhi (ST), and Salmonella enterica serotype Paratyphi A
(SPA), B and C respectively. Global estimates in 2000 suggested 21 million illnesses and
>210,000 deaths due to typhoid fever, plus 5.4 million cases of paratyphoid fever, with
similar figures also estimated for 2010 [1,2]. Enteric fever is endemic in Nepal with seasonal
us
outbreaks; however studies suggest that sustained higher incidence of SPA cases is being
observed in recent years [3-5]. It is among the commonest causes for adults to visit health
an
and SPA have been reported to cause indistinguishable clinical syndromes and disease
severity [7]. Moderately effective vaccines against ST are available, but vaccine uptake was
negligible in Nepal before 2011 [8]. Previously, a randomized controlled trial of Vi-based
vaccine in Nepal was shown to be 75% effective against ST, but it does not confer protection
pt
ed
Discovery of chloramphenicol in 1948, and its use in treatment for enteric fever decreased
case-fatality due to infection from 20% to <1% [10]. Antimicrobials such as chloramphenicol,
ampicillin and co-trimoxazole were the drugs of choice for treatment of typhoid until the late
ce
1980s. Since then, the increasing incidence of multidrug-resistant (MDR) ST strains (i.e.
resistant to chloramphenicol, ampicillin and co-trimoxazole) led to the use of
fluoroquinolones (FQs), particularly ciprofloxacin for empiric treatment of enteric fever
Ac
[11,12]. However, resistance to fluroquinolones quickly emerged, as has been reported from
South Asia and elsewhere since early 1990s [13-15].
Understanding the evolving epidemiology and drug resistance can guide policy makers
regarding appropriate interventions and optimal empiric treatment. Previously, there have
facilities, with estimated incidence of >100 cases per 100,000 population [1,6]. In Nepal, ST
been prevalence studies at different time points; however a sufficiently long-term prospective
surveillance or time-trend analysis, particularly describing ST and SPA separately, has not
cr
ipt
been published. Thus, we conducted a systematic review and meta-analysis to understand the
trends in etiology, and antimicrobial resistance to ST, and SPA in Nepal from 19932011using available evidence.
METHODS
an
antimicrobial use and policy is likely to be relatively homogeneous. The Kathmandu Valley,
divided into three major districts, is densely populated (2.5 million), where most enteric fever
cases are reported [16,17] and where most studies included in this review were conducted.
Provision of clean water and sanitation remain huge problems, with faecal coliform
pt
ed
contamination in 100% of surface water and 60% of piped water at distribution points [18].
The majority of drug sales (~90%) occurs through private pharmacy retailers, who are often
untrained [19]. This has led to indiscriminate and unnecessary use of antimicrobials for a
range of illnesses [20]. Although antimicrobials are legislatively mandated as prescription
ce
Ac
We searched in PubMed, MEDLINE, EMBASE, Google Scholar and the thesis directory of
Tribhuvan University of Nepal for relevant literature published/reported prior to December
2012. An electronic search was conducted using the following terms: enteric fever,
salmonella, typhoid, paratyphoid, nepal, kathmandu. A manual search was
conducted on all relevant references listed within articles identified after an initial screen. We
us
Settings
included data from unpublished thesis works submitted to Tribhuvan University, and relevant
topics were identified from the thesis directory. We also performed individual searches in all
cr
ipt
peer reviewed medical journals published from Nepal (A detailed search strategy is listed in
Appendix 1).
us
Eligibility criteria
The review included all identified hospital-based studies or publications from the national
an
cases of enteric fever. Studies were included if antimicrobial resistance was measured by
standard disk-diffusion method according to guidelines from the Clinical and Laboratory
Standard Institute (CLSI; formerly NCCLS). Data from submitted theses to Tribhuvan
University, which are peer reviewed, were included if sufficient details were reported to meet
pt
ed
eligibility criteria.
Studies were excluded if the majority of cases were from single source outbreaks, as they can
disproportionally distort trends. Studies with combined reporting of ST and SPA
antimicrobial susceptibility data were excluded, as were studies reporting antimicrobial
ce
susceptibility on isolates from stool culture. The study selection process is shown in Figure 1.
Quality of studies and publication bias
Ac
We measured the quality of studies using a context-modified version of a tool generated for
assessing study quality on prevalence [21]. According to this tool, studies with a minimal risk
of bias have a maximum total quality score (TQS) of 19. The criteria for assessment of
quality are based on sampling, measurement and analysis. The outcome measurements in our
study involved laboratory methods rather than survey questionnaire, thus the assessment tool
surveillance system reporting proportions of ST and SPA among all blood culture-confirmed
was adapted accordingly by replacing the term survey questionnaire with laboratory
methods. As blood culture and antimicrobial susceptibility testing are relatively standard
cr
ipt
procedures in hospital laboratories, most studies attained good scores. However all studies
were based on convenience sampling and thus scored poorly on sampling criteria. We
categorized studies in low (TQS, 17-19), moderate (TQS, 14-16) and high risk of bias (TQS,
11-13), arbitrarily. Studies with a TQS of <11 were excluded.
us
To examine whether there was evidence for publication or reporting biases, particularly
whether studies reporting higher resistance rates were more likely to be published, we
an
addition, we carefully assessed for the duplication of cases between studies by an approach
similar to that suggested by Wood [22] where studies were examined to check if they were
published by the same authors, from the same hospital/institution, or where data was
collected during the same time period. The corresponding author was contacted if further
pt
ed
Two authors (S.K and P.S) extracted data independently using a standard data collection form.
Information was compiled on total culture-confirmed enteric fever, exact year and location of
ce
sampling, number of ST and SPA isolates, number of MDR ST and SPA isolates, total
number of isolates tested for individual antimicrobials and number susceptible, intermediate
Ac
or resistant to each antimicrobial. MDR was defined as resistance to three or more drugs
(chloramphenicol, co-trimoxazole, ampicillin or amoxicillin). All studies used Kirby-Bauer
disc diffusion method to measure susceptibility and a few additionally used minimum
inhibitory concentrations for further analyses; however proportion of resistance was
determined from results of disc diffusion. Resistance to nalidixic acid, chloramphenicol, co-
performed visual analysis with funnel plot and statistical analysis with Eggers test. In
trimoxazole, amoxicillin and ampicillin was defined as isolates having a zone of inhibition
within the category of resistant according to CLSI guidelines. However, for public health
cr
ipt
purposes, CLSI emphasizes the importance of reporting both intermediate and resistant
isolates of Salmonella for fluroquinolones and third generation cephalosporins [23]. Thus, we
included intermediate and resistant strains in one non-susceptible category for
ciprofloxacin and ceftriaxone. Two large studies reported data from multiple years with
us
antimicrobial resistance patterns and etiology of enteric fever broken down by year, thus we
included the annual data of these studies as different data-sets. Where studies reported
an
Statistical analysis
regression model. The outcome of meta-regression model was the odds ratio. Firstly, the
proportion (p) (proportion of etiology (ST or SPA), and proportion of resistance to individual
pt
ed
antimicrobials) values were transformed to logit (log of the odds). A linear regression of
logit transformed values over time (year) was performed giving the following relationship
between the proportion , and time .
ce
The analytical weight was proportional to sample size, and we expressed the result of
regression as odds ratio (exponentiation of coefficient () gives odds ratio). The odds ratio
Ac
from our model can be interpreted as the relative change in odds (i.e odds of ST, odds of SPA,
or odds of resistance to specific antimicrobials) per year.
Additionally, if the relative change in odds of etiology or resistance was not significant, we
calculated the overall pooled proportion. The reported proportion of resistance to individual
summary data from multiple years we used the mid-interval year as the study time point.
antimicrobials in most studies was often either near 0 or near 1, and some studies reported no
resistant cases. To stabilize the variance in such studies, we used Freeman-Tukey type double
(where
cr
ipt
tested) [24]. In studies where resistance was not detected in any isolate, the proportion was
estimated as
us
analysis using DerSimonian-Laird weighting. The inferential values were back transformed
from the pooled estimate of transformed values [25]. Preliminary meta-regression showed
an
RESULTS
pt
ed
Study characteristics
Out of the total 32 studies involving investigation of 21,067 ST and SPA isolates, only 5 were
reported from outside of the Kathmandu Valley and only 6 were conducted in children under
age 15 years. The pooled proportion of Salmonella isolation among all blood cultures from
ce
1993-2011 was 9.2% (95% CI 7.4, 11.0), with no significant trend over time (p=0.07). The
sample size, year of sampling, and proportions of SPA isolates among Salmonella isolates are
Ac
shown in Figure 2.
The score for quality of studies varied from 13 to 17 on the assessment scale. Out of 32
studies, 4 were categorized as having low risk of bias, 24 as moderate risk and 4 as high risk.
Since there were few studies with low or high risk of bias, sensitivity analyses were not
performed based on study quality. In the funnel plot analysis, we observed considerable
that resistance varied with time but not with geographical location, thus we calculated the
cr
ipt
confirmed enteric fever cases [4-7,26-49]. The pooled proportion of enteric fever due to SPA
an
increased from 25.0% in years 1993-1997 to 41.0% in years 2008-2011 (Figure 2). The
relative change in odds of SPA as an etiology of enteric fever increased significantly over
time [Odds ratio (OR) = 1.04, 95% CI 1.02, 1.06]. The visual illustration of the decreasing
pt
ed
ce
compared to ST (Table 1) (Figure 3.B). The pooled prevalence of MDR ST was 3.0% (95%
CI 0.3, 8.1) and MDR SPA was 0.5% (95% CI 0.01, 1.7) for 2008-2011 (Table 1).
Ac
us
Twenty-eight studies reported the proportion of ST and SPA isolates among all blood-culture
10
however the relative change in odds of NA in SPA sharply increased over time (OR= 1.24,
95% CI 1.08, 1.41) (p<0.001), and the pooled prevalence was 90.5% (95% CI 85.6, 95.4) in
cr
ipt
us
intermediate resistance) significantly increased over time for both ST (p<0.01) and SPA
(p<0.05). In the years 2008-2011, the pooled prevalence of ciprofloxacin resistance in ST and
an
pt
ed
prevalence of ceftriaxone was 1.5% in ST and 0.7% in SPA (Table 1) (Figure 4.E and 4.F).
Trends in resistance to chloramphenicol, and co-trimoxazole
Nineteen studies reported chloramphenicol resistance [5,7,26-28,32,34,35,39,42-44,46,47,4952,54], and eighteen studies reported co-trimoxazole resistance [5-7,26-28,32,34,35,39,42-
ce
Ac
SPA was 10.65% (95% CI 6.3, 15.8) and 14.3% (95% CI 5.2, 26.9), respectively (Table 1)
11
resistance to chloramphenicol, and co-trimoxazole for SPA was 3.2% (95% CI 2.0, 4.6) and
2.0% (95% CI 1.1, 3.2), respectively (Table 1).
cr
ipt
DISCUSSION
intermediate resistance) is significantly increasing for ST and SPA, but ceftriaxone resistance
an
has remained relatively low for both. Conversely, the prevalence of resistance to traditional
first line antibiotics like, chloramphenicol, and co-trimoxazole is significantly decreasing.
The prevalence of MDR ST and MDR SPA is also decreasing, reflecting the decreasing
The selective pressures behind the shifting etiology of enteric fever are not well understood.
pt
ed
ce
Ac
us
increasing sharply in SPA, but has remained stable in ST. Ciprofloxacin resistance (including
12
equally throughout the year whereas ST cases peak in July and August, the months with
highest annual precipitation, again possibly reflecting differential exposure risks [4,5].
cr
ipt
The fact that SPA is increasingly being common simultaneously with steep increases in NA
and ciprofloxacin resistance raises serious concerns about empirical treatment of suspected
us
an
isolates, a novel mutation in quinolone resistance determining region has been reported from
Nepal [59]. A linear increasing trend in the MIC of ciprofloxacin over time has also been
As has been reported from other countries in South-Asia [60], we observed a low and
pt
ed
ce
MDR isolates was observed, probably due to reduced use of these antibiotics over recent
decades. Similar results have been reported from India [61,62]. We speculate that the
Ac
restoration of susceptibility may have a genetic basis with resistance residing in plasmids.
These findings suggest that there is an opportunity to re-consider antibiotics which were
rendered non-effective by resistance 20 years previously. Alternatively, there are a few other
drugs like gatifloxacin and azithromycin whose efficacy for the treatment of enteric fever has
been tested recently in clinical trials [63,64]. However, clinicians involved in treating patients
13
with enteric fever should consider various factors like cost, availability, dose requirements,
and patient co-morbidities in relation to side effects, before making specific antimicrobial
cr
ipt
choices.
Our study has several limitations. Although we calculated the predicted relative change in
odds of etiology/drug-resistance patterns, findings are solely for understanding trends over
time and should not be used for prediction beyond the time period of study or to predict
us
clinical efficacy. Additionally, the current CLSI recommended zone of inhibition for
fluoroquinolone susceptibility is controversial and may not correlate with true in vivo
an
of resistance to NA, and/or preferably on the basis of MICs for individual fluroquinolones.
Further, we do not have information about the extent of clustering of cases of enteric fever in
the studies included in the meta-analysis, as most of the primary studies did not explore for
such an observation. However, it should be noted that when the cases are clustered in
biased.
pt
ed
outbreaks, they do not supply independent information and the estimate of prevalence may be
ce
Ac
effectiveness. The functional susceptibility for fluroquinolones should be judged on the basis
14
NOTES
Acknowledgements
cr
ipt
We thank Professor John Crump, University of Otago, New Zealand for comments on a draft
of this manuscript, and Associate Professor Rory Wolfe, Monash University, Australia for
Contributors
an
SK and AC designed the study. SK and PS extracted the data, assessed the quality of studies,
and independently decided for inclusion or exclusion, and in events of disagreement, SD
helped to resolve. SD contacted corresponding authors of potentially eligible studies, if
needed. SK performed all the statistical analyses with help from AC, and KL. SK prepared
manuscript with the help from AC, and KL. All authors contributed to the draft of this report
Funding
None
pt
ed
ce
Conflict of interests
Ac
us
statistical advice.
15
References
1.
Crump JA, Luby SP, Mintz ED. The global burden of typhoid fever. Bull World
2.
cr
ipt
Buckle GC, Walker CL, Black RE. Typhoid fever and paratyphoid fever: Systematic
review to estimate global morbidity and mortality for 2010. J Glob Health 2012; 2(1):
10401.
us
3.
point source, multidrug-resistant outbreak in Nepal. Clin Infect Dis 2005; 40(4): 554-
an
4.
Karkey A, Arjyal A, Anders KL, et al. The burden and characteristics of enteric fever
at a healthcare facility in a densely populated area of Kathmandu. PLoS One 2010;
5.
5(11): e13988.
Acharya D, Bhatta DR, Malla S, Dumre SP, Adhikari N, Kandel BP. Salmonella
pt
ed
Murdoch DR, Woods CW, Zimmerman MD, et al. The etiology of febrile illness in
adults presenting to Patan hospital in Kathmandu, Nepal. Am J Trop Med Hyg 2004;
70(6): 670-5.
Maskey AP, Day JN, Phung QT, et al. Salmonella enterica serovar Paratyphi A and S.
ce
7.
Ac
8.
Balayla J, Azoulay L, Abenhaim HA. Maternal marital status and the risk of stillbirth
and infant death: a population-based cohort study on 40 million births in the United
States. Womens Health Issues 2011; 21(5): 361-5.
61.
16
9.
Acharya IL, Lowe CU, Thapa R, et al. Prevention of typhoid fever in Nepal with the
Vi capsular polysaccharide of Salmonella typhi. A preliminary report. N Engl J Med
10.
cr
ipt
chloromycetin in the treatment of typhoid fever. Ann Intern Med 1948; 29(1): 131-4.
11.
Bhan MK, Bahl R, Bhatnagar S. Typhoid and paratyphoid fever. Lancet 2005;
12.
us
366(9487): 749-62.
Bhutta ZA. Current concepts in the diagnosis and treatment of typhoid fever. BMJ
an
13.
Launay O, Nguyen Van JC, Buu-Hoi A, Acar JF. Typhoid fever due to a Salmonella
14.
Parry C, Wain J, Chinh NT, Vinh H, Farrar JJ. Quinolone-resistant Salmonella typhi
pt
ed
15.
ce
17.
Ac
18.
17
http://web.mit.edu/watsan/Docs/Student%20Theses/Nepal/Wolfe2000.pdf, accessed
23 January 2013], 2000.
Joshi MP, Khakurel B. Drug rationalization: now for the hard part. World Health
cr
ipt
19.
us
21.
for assessing the quality of studies on prevalence. Clin Oral Investig 2012; 16(3): 781-
an
22.
Stacey T, Thompson JM, Mitchell EA, Ekeroma AJ, Zuccollo JM, McCowan LM.
23.
The Auckland Stillbirth study, a case-control study exploring modifiable risk factors
for third trimester stillbirth: methods and rationale. Aust N Z J Obstet Gynaecol 2011;
24.
pt
ed
51(1): 3-8.
Miller JJ. The Inverse of the Freeman-Tukey Double Arcsine Transformation. The
ce
25.
Maskey AP, Basnyat B, Thwaites GE, Campbell JI, Farrar JJ, Zimmerman MD.
Emerging trends in enteric fever in Nepal: 9124 cases confirmed by blood culture
Ac
26.
8.
18
27.
Shrestha B. Isolation of Salmonella spp. From blood and study of its antibiotic
sensitivity pattern: A master's thesis sumbitted to Tribhuvan University, Kathmandu,
28.
cr
ipt
Nepal, 1996.
Shakya R. Study on bacterial flora in blood specimen collected from hospitalized and
out patients services of T.U Teaching Hospital: A master's thesis submittted to
Tribhuvan University, Kathmandu, Nepal, 2001.
us
29.
typhoid and paratyphoid fever in Kathmandu: two years study and trends of
Guha S, Jalan BY, Dey S, Easow JM, Wilson G, Shivananda PG. Salmonella
an
30.
31.
7(1): 23-5.
pt
ed
Shrestha S. Study on the prevalence of Salmonella species from blood sample of the
patients visiting National Public Health Laboratory, Teku: A master's thesis
ce
Pokharel BM, Koirala J, Dahal RK, Mishra SK, Khadga PK, Tuladhar NR.
Multidrug-resistant and extended-spectrum beta-lactamase (ESBL)-producing
Ac
Salmonella enterica (serotypes Typhi and Paratyphi A) from blood isolates in Nepal:
surveillance of resistance and a search for newer alternatives. Int J Infect Dis 2006;
10(6): 434-8.
19
34.
35.
cr
ipt
us
36.
enterica serovar Typhi associated with a class 1 integron carrying the dfrA7 gene
an
37.
38.
39.
pt
ed
ce
40.
presenting to a general hospital outpatient clinic in urban Nepal. PLoS One 2012;
7(10): e47531.
Mishra SK, Sah JP, Shrestha R, Lakhey M. Emergence of nalidixic acid resistant
Ac
41.
20
42.
43.
cr
ipt
serovars from enteric fever suspected patients visiting Bir hospital: A master's thesis
submitte to Tribhuvan University, Kathmandu, Nepal, 2009.
Prajapati B, Rai GK, Rai SK, et al. Prevalence of Salmonella typhi and paratyphi
us
44.
infection in children: a hospital based study. Nepal Med Coll J 2008; 10(4): 238-41.
Pokharel P, Rai SK, Karki G, Katuwal A, Vitrakoti R, Shrestha SK. Study of enteric
an
Joshi BG, Keyal K, Pandey R, Shreshta BM. Clinical Profile and Sensitivity Pattern
46.
pt
ed
47.
Salmonella enteric serovar Typhi and Paratyphi A from blood samples of suspected
enteric fever patients. International Journal of Infection and Microbiology 2012; 1(1):
9-13.
ce
48.
among the patients attending Shree Birendra Hospital. Medical Journal of Shree
Birendra Hospital 2011; 10(2): 1-3.
Ac
49.
45.
21
50.
51.
cr
ipt
Enteric Fever in Nepali Children. Post graduate Medical Journal of NAMS 2008; 8(1):
1-4.
Singh UK, Neopane AK, Thapa M, Aryal N, Agarwal K. Salmonella Typhi Infections
us
52.
an
53.
54.
pt
ed
enterica serovar Typhi isolated from Dhulikhel Hospital, Nepal. Jpn J Infect Dis 2012;
65(3): 264-7.
55.
ce
56.
Vollaard AM, Ali S, van Asten HA, et al. Risk factors for typhoid and paratyphoid
Ac
57.
22
cr
ipt
58.
susceptibility in Salmonella enterica isolates from travelers, Finland. Emerg Infect Dis
2009; 15(5): 809-12.
59.
Koirala KD, Thanh DP, Thapa SD, et al. Highly resistant Salmonella enterica serovar
us
Typhi with a novel gyrA mutation raises questions about the long-term efficacy of
older fluoroquinolones for treating typhoid fever. Antimicrob Agents Chemother 2012;
an
60.
enterica serotype Typhi strains isolated over a period of two decades, with special
reference to ciprofloxacin and ceftriaxone. Int J Antimicrob Agents 2010; 35(5): 4825.
61.
pt
ed
Dutta S, Sur D, Manna B, Bhattacharya SK, Deen JL, Clemens JD. Rollback of
Salmonella enterica serotype Typhi resistance to chloramphenicol and other
ce
Ac
2(6): e542.
64.
Butler T. Treatment of typhoid fever in the 21st century: promises and shortcomings.
56(5): 2761-2.
an
M
Figure 1: Flow diagram for study selection process (ST= Salmonella enterica serotype Typhi,
Ac
ce
pt
ed
us
cr
ipt
23
an
M
pt
ed
Figure 2: Forest plot showing pooled proportion of enteric fever due to Salmonella enterica
ce
Ac
us
cr
ipt
24
an
M
Figure 3: (A) Proportion of Salmonella enterica serotype Typhi (ST) among total Salmonella
pt
ed
enterica isolates (p<0.01, for trend). (B) Proportion of multi-drug resistant ST among total
ST isolates. Each grey circle indicates an individual study, and area of circle size is
proportional to the sample size. The sizes of circles are comparable within individual graphs
Ac
ce
only. The trend line (dashed) is the predicted prevalence from the regression model.
us
cr
ipt
25
an
M
pt
ed
Salmonella enterica serotype Typhi (ST). (B) Nalidixic acid resistance in Salmonella enterica
serotype Paratyphi A (SPA). (C) Ciprofloxacin resistance in ST. (D) Ciprofloxacin resistance
in SPA. (E) Ceftriaxone resistance in ST. (F) Ceftriaxone resistance in SPA. Each grey circle
indicates an individual study, and the area of circle size is proportional to the sample size.
ce
The sizes of circles are comparable within individual graphs only. The trend line (dashed) is
the predicted prevalence from the regression model. [Please note the prevalence scales are
Ac
us
cr
ipt
26
an
M
pt
ed
Ac
ce
regression model. [Please note the prevalence scales are unique to each figure]
us
cr
ipt
27
28
MDR
SPA
ST
SPA
ST
Year (1998-
Year (2003-
Year (2008-
Overall
1997)
2002)
2007)
2011)
(1993-2011)
NA
10.3 (0.7,
6.5 (3.3,
3.0 (0.3,
28.5)
10.7)
8.1)
NA
2.2 (1.1,
0.5 (0.01,
3.6)
1.7)
NA
NA
NA
NA
NA
NA
NA
1.6 (0.04,
90.5 (85.6,
91.7)
95.4)
2.5 (0.6,
10.6 (6.3,
Ceftriaxone
NA
5.3)
5.8)
15.8)
3.9 (0.1,
2.9 (0.4,
14.3 (5.2,
pt
ed
SPA
ST
NA
ce
SPA
NA
Chloramphenicol
ST
Ac
SPA
Co-trimoxazole
ST
SPA
NA
NA
NA
83.9 (76.0,
Ciprofloxacin
Year (1993-
16.2)
7.5)
26.9)
NA
NA
NA
NA
NA
NA
(95% CI)
0.87 (0.78,
0.97)**
NA
0.87 (0.81,
0.95)**
63.4 (56.2,
0.97 (0.85,
70.7)
1.11)
NA
1.24 (1.08,
1.41)**
NA
1.14 (1.04,
1.26)**
NA
1.11 (1.01,
1.22)*
1.22 (0.94,
1.58)
NA
0.95 (0.88,
1.02)
13.6 (5.1,
4.7 (0.5,
4.2 (1.9,
25.2)
12.9)
7.2)
NA
1.2 (0.2,
3.2 (2.0,
2.8)
4.6)
19.1 (10.5,
12.5 (6.0,
2.4 (0.9,
6.9 (3.4,
29.6)
20.9)
4.6)
11.6)
4.9 (3.6,
1.7 (1.2,
1.2 (0.3,
2.0 (1.1,
6.4)
2.4)
2.8)
3.2)
NA
NA
0.89 (0.81,
0.99)*
NA
0.85 (0.82,
0.89)***
NA
Nalidixic acid
Odds ratio
cr
ipt
ST
us
MDR
Isolate
an
Antimicrobials
29
cr
ipt
serotype Paratyphi A. Pooled proportion in five-year time blocks is not shown if there were
less than 3 studies within that year block, and is noted as NA. Overall pooled proportion was
not shown if the relative change in odds is significant over time (1993-2011), and is noted as
NA. The odds ratio (OR) can be interpreted as relative change in odds per year. A
an
M
pt
ed
ce
Ac
us