Accelerat ing t he world's research.

Where do developing world clinicians

obtain evidence for practice
Murtaza Akhtar

Journal of Clinical Epidemiology

Cite this paper Downloaded from Academia.edu 

Get the citation in MLA, APA, or Chicago styles

Related papers Download a PDF Pack of t he best relat ed papers 

At t it udes of developing world physicians t o where medical research is performed and report ed
William Macharia

Burden of t uberculosis in nine sub-Saharan African count ries

Tolulope Balogun

Are all eligible t uberculosis pat ient s in nat ional t uberculosis programme screened for drug resist ance…
Deepak Shewade
 Journal of Clinical Epidemiology 53 (2000) 669–675

Where do developing world clinicians obtain evidence for practice:

A case study on pneumonia
John Pagea, Richard F. Hellera,*, Scott Kinlaya, Lynette L.-Y. Lima, Wang Qianb,
Zheng Supingc, Supornchai Kongpatanakuld, Murtaza Akhtare,
Salah Khedrf, and William Machariag
a
Centre for Clinical Epidemiology and Biostatistics; The University of Newcastle, Newcastle, Australia;
b
Clinical Epidemiology Unit, Shanghai Medical University, Shanghai, China;
c
Department of Clinical Epidemiology, West China University of Medical Sciences, Chengdu, China;
d
Department of Pharmacology, Faculty of Medicine, Siriraj Hospital, Mahidol University, Bangkok, Thailand;
e
Government Medical College, Maharastra State, Nagpur, India;
f
Clinical Epidemiology Unit, Suez Canal University, Suez Canal, Ismalia, Egypt;
g
Clinical Epidemiology Unit, Faculty of Medicine, University of Nairobi, Nairobi, Kenya
Received 18 January; accepted 22 December 1999

Abstract
There are few data on the practice of evidence based medicine in the developing world, nor on the actual sources of evidence that cli-
nicians use in practice. To test the hypothesis that there was variation between and within developing countries in the proposed manage-
ment of a patient with hospital acquired pneumonia, and that part of the variation can be explained by the sources of evidence used.
Questionnaire responses to hypothetical case history. Investigators from 6 centres within the International Clinical Epidemiology Net-
work (INCLEN) in China, Thailand, India, Egypt, and Kenya. Doctors chosen to represent primary and secondary hospital practice in
the regions of the study centres. Investigations and initial treatments which would be ordered for a hypothetical 60-year-old woman who
develops pneumonia 5 days after hospital admission, whether local data on antibiotic sensitivities are available and where information
would be obtained to guide management. Chest x-ray and sputum gram stain/culture were consistently the most commonly ordered inves-
tigations, there being much greater variation in the initial treatment choices with either penicillin, a third-generation cephalosporin or
aminoglycoside being the most popular choice. Textbooks were the commonest form of information source, and access to a library, text-
books and journals were statistically significantly associated with appropriate choice of investigations, but not treatment. Access to local
antibiotic sensitivities was associated with appropriate initial treatment choice. Improving access to information in the literature and to lo-
cal data may increase the practice of evidence-based medicine in the developing world. © 2000 Elsevier Science Inc. All rights reserved.
Keywords: Clinical research; Developing countries; Evidence based medicine; Physician’s practice pattern; Pneumonia; Questionnaires

1. Introduction cerned, was evidence based [2]. There is also some evidence
that this also true of inpatient surgical practice in the UK [3].
There is increasing recent debate about the way in which
There is, however, widespread variation in medical practice
evidence should inform clinical practice [1]. A survey from
between and within countries even in areas where good evi-
the UK suggested that a large proportion of hospital-based
dence exists [4]. There are few data on the practice of evi-
clinical decision making, at least so far as treatments are con-
dence-based medicine in the developing world where con-
straints to implementing effective health care may include a
lack of access to literature and to trial evidence from local
* Corresponding author. Centre for Clinical Epidemiology and Biosta- populations, lack of financial resources, lack of regulation of
tistics, Faculty of Medicine and Health Sciences, The University of New- pharmaceutical companies, and politics [5]. In addition, there
castle, Level 3, David Maddison Clinical Sciences Building, Royal
Newcastle Hospital, Cnr King and Watt Streets, Newcastle, NSW 2300,
is little study of the actual sources of evidence that clinicians
Australia. Telephone: 161 2 49236142; Fax: 161 2 49236148. use in practice. We designed a study to test the hypothesis
E-mail address: rfhcceb@ibm.net that there was variation between and within countries in the
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved.
PII: S0895-4356(99)00 2 3 1 - 0
670 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675

proposed management of a patient with hospital acquired Questionnaires or discs with coded data were sent to the
pneumonia, and that part of the variation can be explained by coordinating centre in Newcastle, where analyses were per-
the sources of evidence used to inform clinical decision mak- formed.
ing. This involved a survey of hospital doctors working in six On the basis of a consensus among the investigators, we
regions of five countries in the developing world. decided that there was good evidence to support the choice
of a full blood count, a chest X-ray, sputum gram stain/cul-
ture, and blood cultures for investigations and either a
2. Methods
parenteral third-generation cephalosporin, extended spec-
Clinicians who are members of the International Clinical trum penicillin, aminoglycoside or a quinolone as initial
Epidemiology Network (INCLEN) were invited to partici- treatment (before test results were available). These were
pate, and six centres agreed: Shanghai and Chengdu in China, deemed “appropriate” investigation and treatment, respec-
Bangkok in Thailand, Nagpur in India, Ismalia in Egypt, tively.
and Nairobi in Kenya. The study was coordinated in the
Centre for Clinical Epidemiology and Biostatistics in New- 3. Statistical methods
castle, Australia.
Each centre was asked to identify hospitals and physi- The data were tabulated and logistic regression was used
cians within those hospitals who would be expected to treat to calculate unadjusted odds ratios and P values. These were
patients with pneumonia, in a way that would represent the done by using the statistical program Stata version 4.0.
generality of tertiary and secondary hospital settings in their Multilevel statistical models were fitted to the data with
region. A questionnaire was given to each consenting doc- the aid of the statistical program Mln [6]. Individuals were
tor. The sampling procedure varied between centres. In modelled as level 1 units clustered within hospitals as level
Bangkok, Chengdu, Shanghai, and Nagpur a sample of ter- 2 units which were in turn clustered within centres as level 3
tiary care (3 in Bangkok and Nagpur, 5 in Chengdu and units. Logistic regression models were fitted to the data with
Shanghai), and secondary care (4 in Bangkok and 5 in appropriate investigations and appropriate treatment as out-
Chengdu, Shanghai, and Nagpur) hospitals were selected ei- come variables respectively. Statistical significance was de-
ther at random or to cover the spread of teaching/non-teach- termined by P values after calculating deviances and com-
ing, geography and hospital size. In Ismailia, all three city paring the changes with the appropriate chi-square (x2)
hospitals (one tertiary and two secondary) were chosen. In distribution as variables were subtracted from the models.
Nairobi, a list of all the physicians in the country rather than The variables investigated include physician specialty, sex,
hospitals was the sampling frame. access to a medical library, hospital specialisation, rural
Physicians within the hospitals were chosen from those versus urban/suburban location, stated use of journal infor-
working in internal medicine either at random or to repre- mation/computer search, local journals, Western (North
sent a spread of academic/non-academic and seniority lev- American/European) journals, local textbooks, Western
els. In Bangkok there were 25 doctors from each of the ter- textbooks, country of practice, physician’s age, and number
tiary and secondary groups of hospitals, in Chengdu and of years since graduation. Variables remained in the model
Shanghai there were 5 from each hospital. In Nagpur, a ran- if they were statistically significantly related to the appro-
dom sample was taken to produce 5 doctors in the tertiary priate choice of investigations or treatment (P , 0.05). All
care hospitals and in the secondary hospitals all doctors variables were adjusted for this set of variables in multiple
were asked to participate. In Ismalia, a random sample was regression analysis (Tables 1 and 2). However, stated use of
taken of all relevant doctors in the three hospitals. any textbook and use of Western and local textbooks were
The questionnaire started with a hypothetical case his- not included in the same models due to collinearity. The
tory, representing a 60-year-old patient admitted to hospital same applies to stated use of any journals/computer search
with a hip fracture who develops pneumonia after one week and use of Western and local journals.
(Appendix A). Respondents were asked to indicate which In cases where the statistical program computed devi-
investigations they would order and what treatment they ances that would result in negative values for change, the
would prescribe initially (before test results were available). latter was truncated to zero (0), and the P value reported as
The questionnaire then asked what was the source of the in- one (1). Since none of these variables remained in the final
formation used to inform the patient’s management (from a model, interpretation was unaffected.
list of potential sources provided), whether local data on fre- Further details on the use of multilevel modelling are de-
quencies of micro-organisms and antibiotic sensitivities are scribed by Byrk and Raudenbush [7]. All P values are two
available, how they rate research in journals from North sided.
America, Europe, their region or country as being likely to
influence their clinical practice and what is their usual
4. Results
source of up to date clinical information. A few demo-
graphic questions concluded the questionnaire. One re- Responses to the survey were mostly high. In both cen-
minder was sent to non-respondents. tres in China and in Nagpur, all the contacted doctors sent a
 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675 671

Table 1
Use of appropriate investigations*
Appropriate
investigation Multilevel
Variable (n/N) % Univariate OR P value multivariate OR P value
Specialty
Primary care physicians 19/71 27 1 0.02 1 0.004
Pulmonary specialists 12/39 31 1.3 1.1
Other specialist physicians 20/79 25 1.0 1.0
Other doctors 1/17 6 0.2 0.3
Sex
Males 31/123 25 1 0.94 1 0.82
Females 21/85 25 1.0 0.8
Access to a medical library
No 3/24 13 1.0 0.11 1 0.18
Yes 49/184 27 2.5 1.81
Hospital specialisation
Secondary hospitals 18/75 24 1 0.30 1 0.46
Tertiary hospitals 29/93 31 1.4 1.55
Hospital location
Rural hospitals 9/25 36 1 0.34 1 0.27
Urban/suburban hospitals 38/143 27 0.6 0.73
Use of journal information/computer search
No 12/76 16 1 0.002 1 0.011
Yes 40/132 30 2.3 1.7
Information from local journals
No 19/102 18 1 0.04 1 0.41
Yes 33/106 32 2.0 1.2
Information from Western journals*
No 36/141 26 1 0.92 1 0.46
Yes 16/67 24 1.0 1.1
Information from textbooks
No 0/17 0 1
Yes 55/191 27 ∞
Use of local textbooks
No 5/53 9 1 0.001 1 ,0.001
Yes 47/155 30 4.2 4.1
Use of Western textbooks
No 24/113 21 1 0.17 1 ,0.001
Yes 28/95 29 1.6 2.3
Study sites
Chinese centres 30/100 30 1 0.091 1 0.92
Bangkok 12/40 30 1.0 0.68
Nagpur 5/28 18 0.51 0.36
Nairobi 5/40 13 0.33 0.56
*Minimum of CBC, chest Xray, sputum gram stain/culture and blood cultures. Unless stated otherwise, percentages are in brackets.

response. In Bangkok, all of the doctors from the tertiary jority treated between 5 and 25 cases per year) and greater
hospitals responded, as did 15 of the 25 (60%) of those in than 75% had access to a medical library.
secondary hospitals. In Ismailia, 20 of 22 (91%) of the doc- Table 4 shows where physicians would obtain informa-
tors responded. In Nairobi, however, only 40 of the 84 tion for managing such a patient with hospital acquired
(48%) doctors replied. pneumonia. Textbooks were the most common source, jour-
Table 3 shows the major demographic and practice fea- nals less popular and computer searching uncommon. Local
tures of the physicians from the six centres that were stud- data on frequency and sensitivity of the organisms were
ied. There were fewer females in the samples from available in varying proportions—from 35 and 35% respec-
Bangkok, Ismalia, and Nairobi compared to the other sam- tively in Nairobi to 83 and 75% respectively in Bangkok.
ples. Compared to the other samples, physicians from The proportions who would perform the most common
Bangkok and Ismalia were more recent graduates from investigations are shown in Table 5. There was reasonable
medical schools. The samples from China were the only consistency here, with chest x-ray and sputum gram stain/
ones that had appreciable numbers of pulmonary specialists. culture being frequently requested, although blood cultures
The majority of doctors treated more than 25 cases of pneu- ranged from 0 to 46% between centres and blood gases be-
monia per year (with the exception of Ismalia where a ma- tween 0 and 22%.
672 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675

Table 2
Use of appropriate treatment*
Multilevel
Variable Appropriate treatment (n/N) % Univariate OR P value multivariate OR P value
Specialty
Primary care physicians 38/71 54 1 0.28 1 0.56
Pulmonary specialists 28/39 72 2.1 0.61
Other specialist physicians 44/79 56 1.0 0.59
Other doctors 9/16 53 0.93 1.1
Sex
Males 69/123 56 1 0.47 1 1
Females 52/85 61 1.2 0.77
Access to a medical library
No 14/24 58 1 0.99 1 0.89
Yes 107/184 58 1.0 0.61
Hospital specialisation
Secondary hospitals 51/75 68 1 0.24 1 1
Tertiary hospitals 55/93 59 0.68 1.1
Hospital location
Rural hospitals 15/25 60 1 0.73 1 ,0.001
Urban/suburban hospitals 91/143 64 1.2 0.19
Use of journal information/computer search
No 43/76 57 1 0.72 1 1
Yes 78/132 59 1.1 0.87
Use of local journals
No 49/102 48 1 0.040 1 0.90
Yes 65/106 61 1.8 1.0
Use of Western journals
No 92/141 65 1 0.003 1 0.62
Yes 29/67 43 0.41 0.68
Information from textbooks
No 7/17 41 1 0.15 1 0.87
Yes 114/191 60 2.1 1.1
Use of local textbooks
No 23/53 43 1 0.03 1 1
Yes 98/155 63 2.2 1.0
Use of Western textbooks
No 80/113 71 1 ,0.001 1 ,0.001
Yes 41/95 43 0.31 0.39
Information available on antibiotic sensitivities
No 49/90 54 1 0.34 1 0.001
Yes 72/118 61 1.3 1.6
Study sites
Chinese centres 76/100 76 1 ,0.001 1 ,0.001
Bangkok 8/40 20 0.08 0.05
Nagpur 22/28 79 1.2 2.6
Nairobi 15/40 35 0.19 0.07
*Third generation cephalosporin, extended spectrum penicillin, aminoglycoside, or quinolone.

Choice of other tests included a test for TB by 2% of pneumonia. Non-internists were the least likely to state this
doctors in Shanghai and 26% in Nairobi, an ECG ranging choice. Those who stated that they would use information
from none to 18% across the study and a smattering of other from medical journals, computer search of the literature or
tests including ESR, bronchoscopy and lung function tests. textbooks were more likely to choose the appropriate inves-
Table 6 shows the projected choice of treatment. There tigations than those who did not state these information
was considerable variation here and a number of other anti- sources.
biotics were selected by doctors in the various centres. Table 2 shows the relationships between stated use of ap-
Table 1 shows the relationships between stated use of ap- propriate treatment (use of a parenteral third-generation
propriate investigations and physicians’ demographic and cephalosporin, extended spectrum penicillin, aminoglyco-
practice features. Pulmonary specialists were the doctors side or quinolone) and physicians’ demographic and prac-
most likely to state that chest radiograph, complete blood tice features. The variable which was most strongly related
count, sputum gram stain/culture, and blood culture should to the choice of appropriate therapy was study centre. The
all be performed in this suspected case of hospital acquired study participants from the Chinese centres and Nagpur
 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675 673

Table 3
Demographic and practice features of the physicians included in the study

Median no. Specialty

No. of Median of years since Physicans No. of pneumonia cases Access to
No. of No. of females age graduation Primary Pulmonary of other Other managed in last year medical
Centre hospitals physicans (%) (range) (range) care specialists subspecialty doctors ,5 5–25 .25 library
Chengdu 10 50 26 (52) 40 (23–60) 15 (1–37) 14 (28) 16 (32) 20 (40) 0() 0 (0) 13 (26) 37 (74) 49 (98)
Shanghai 10 50 28 (56) 40 (22–64) 14.5 (1–40) 1 (2) 20 (40) 26 (52) 3 (6) 0 (0) 17 (34) 33 (50)
Bangkok 7 40 10 (25) 28 (24–49) 4.5 (1–24) 25 (63) 1 (3) 14 (35) 0 (0) 3 (8) 14 (35) 23 (58) 37 (93)
Nagpur 8 28 11 (39) 38 (26–51) 15 (2–30) 17 (61) 1 (4) 3 (11) 5 (18) 3 (11) 6 (21) 19 (68) 24 (86)
Ismalia 3 20 1 (5) 28 (26–47) 4 (3–23) 3 (16) 0 (0) 16 (84) 0 (0) 1 (5) 13 (68) 5 (26) 15 (75)
Nairobi § 40 10 (25) 40 (34–51) 14 (10–36) 11 (28) 1 (3) 16 (40) 12 (30) 0 (0) 3 (8) 37 (93) 30 (78)
§
The physicians in Kenya were not selected by hospitals.

were more likely to choose the appropriate therapy com-

5. Discussion
pared to those from Bangkok and Nairobi. Other variables
which appeared to have an impact on choice of therapy The study has sampled doctors from a number of hospi-
were those from a rural hospital location, information on an- tals and settings in selected countries across the “develop-
tibiotic sensitivities and an inverse relationship with use of ing” world. The sampling process was designed to identify
Western (European and North American) textbooks. The doctors who would represent the generality of those who
degree of confounding present is noteworthy. In simple would treat patients similar to those in our case history
univariate analyses, it appeared that being a pulmonary spe- across a range of tertiary and secondary hospital settings in
cialist was positively related to choice of appropriate ther- the different regions represented by our study centres. The
apy. However, this was not so when centres were adjusted response rates, with the exception of Nairobi, were gener-
for. The Chinese centres were the ones where physicians ally good. We do not aim to be able to generalise across the
were more likely to choose the appropriate therapy, but “developing” world, but are probably able to do so for most
were also the ones with the majority of pulmonary physi- of the regions included in the study.
cians in the study (Table 3). In these centres, the pulmonary We have used the case history method to describe varia-
specialists were as likely as other physicians to choose the tions in stated practice in the developing world before [8],
appropriate therapy, but none of the three pulmonary phy- although we recognise that there may be differences be-
sicians from non-Chinese centres chose the appropriate tween actual and stated practice. The results from one case
therapy. history may not be generalised to other aspects of clinical
In this study, physicians’ age and experience (number of practice in the same setting, although we have chosen a
years since medical school graduation) were not statistically common clinical scenario of hospital-acquired infection. It
significantly related to stated choice of appropriate investi- is also possible that some of the variation in practice is due
gations or treatment. to lack of availability or cost of the investigation or treat-

Table 4
Percentages quoting various information sources
Chengdu Shanghai Bangkok Nagpur Ismalia Nairobi
(N 5 50) (N 5 50) (N 5 40) (N 5 28) (N 5 20) (N 5 40)
Textbook
Local 100 94 88 68 25 10
North American 14 20 63 75 75 63
European 0 0 35 50 10 63
Any 100 98 93 93 95 75
Journal
Local 74 56 35 54 15 30
North American 8 24 38 43 25 50
European 0 2 25 32 15 50
Any 76 58 43 64 35 63
Computer search 4 4 30 21 0 8
Ask advice 32 2 8 29 10 15
Local data on organisms available
Frequency 70 46 83 50 60 35
Sensitivity 72 50 75 46 75 35
674 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675

Table 5
Percentages performing the most common investigations
Chengdu Shanghai Bangkok Nagpur Ismalia Nairobi
Tests ordered (n 5 50) (n 5 50) (n 5 40) (n 5 28) (n 5 20) (n 5 40)
Chest X-ray 100 100 100 96 100 92
Full blood count 86 84 98 75 40 68
Sputum gram stain/culture 92 92 83 93 75 77
Blood culture 46 28 33 18 0 21
Blood gasses 6 22 0 4 0 5
Electrolytes 14 22 28 7 0 10
Liver function tests 18 12 0 0 0 3
Appropriate investigations 40 20 30 18 13

ment. Most of the tests we describe in this scenario, how- Our choice of appropriate investigation and initial treat-
ever, are simple and readily available in the tertiary and sec- ment was based on consensus among the investigators, due
ondary hospitals that formed the sampling frame for the to the absence of widely acceptable guidelines for the man-
majority of our study samples. There are large cultural differ- agement of hospital-acquired pneumonia. Guidelines do ex-
ences in drug prescribing patterns, which have been well de- ist, in fact many hospitals produce their own and the Ameri-
scribed in the “developing” world [9], and are likely to have can Thoracic Society has published relevant guidelines for
influenced the choices made by the respondents in this survey. the treatment of hospital acquired pneumonia [10]. Most
Textbooks remain the most commonly quoted source of guidelines are limited to the choice of antibiotics and ex-
information about the management of this condition, and clude investigations. Guidelines for the developing world
with the exception of Kenya, local textbooks were preferred are mostly restricted to the diagnosis and treatment of acute
over those from North America which in turn were pre- respiratory infection in children [11]. Our consensus-based
ferred to European textbooks. Journals were also commonly decision on what constitutes evidence based practice is not
used, with local journals being preferred in China, but else- inconsistent with the U.S. guidelines [10], nor with Western
where North American and local journals shared preference. “expert opinion” [12,13]. However, it is possible that local
European journals were less popular with the exception of conditions would make the choice of initial treatment more
Kenya. Computer searching is still uncommon, and asking appropriate than the consensus from our investigators
the advice of colleagues more common than using the com- would indicate. Accepting that our choice can be criticised,
puter, but less common than going to the textbooks or jour- there remains a large variation in proposed management
nals. The acquisition of information to guide evidence based strategies within and between the countries studied.
practice is still mainly via traditional methods, although it We have shown that appropriate investigations are
should be emphasised that this relates to stated not necessar- strongly associated with access to literature based scientific
ily actual practice. evidence (access to medical library, use of local journals
Local data on the frequency and sensitivities of the or- and local and western textbooks). Appropriate therapeutic
ganisms likely to be causing pneumonia were apparently choices do not appear to be related to access to literature-
available to the majority of doctors, although in three of the based scientific evidence, however local information avail-
six countries, only one half or fewer of the doctors said they able on antibiotic sensitivities was strongly related to appro-
would use these data. priate choice of antibiotics.

Table 6
Treatment choice by centre
Chengdu Shanghai Bangkok Nagpur Ismalia Nairobi
Treatment (n 5 50) (n 5 50) (n 5 40) (n 5 28) (n 5 20) (n 5 40)
Penicillin 4 48 33 29 89
Extended spectrum penicillina 56 28 0 0 8
Augmentin 0 0 0 0 3
3rd generation cephalosporin 30 14 0 25 25 8
Aminoglycoside 6 40 20 71 15 20
Quinolone 10 4 0 0 0
Oxygen 8 8 10 14 5 3
Expectorants 72 78 10 32 8
Antipyretics 2 16 15 14 33
Appropriate treatment 84 68 20 79 38
a
Extended spectrum penicillins in this study used Piperacillin and carbenicillin.
 J. Page et al. / Journal of Clinical Epidemiology 53 (2000) 669–675 675

We suggest that improving access to scientific evidence evidence based. A-Team, Nuffield Department of Clinical Medicine.
—which should include local data and locally written texts Lancet 1995;346:407–10.
[3] Howes N, Chagla L, Thorpe M, MacCulloch P. Surgical Practice is
—may be useful in changing clinical practice towards ap- evidence based. Br J Surg 1997;84(9):1220–3.
propriate investigations and treatment. This is another ex- [4] Hampton J. Evidence-based medicine, practice variations and clinical
ample of the importance of the collection of local data to es- freedom. J Eval Clin Pract 1997;3(2):123–31.
sential national health research [14]. [5] Garner P, Kale R, Dickson R, Dans T, Salinas R. Getting research
findings into practice: implementing research findings in developing
countries. Br Med J 1998;317:531–5.
[6] MLn Software for Multilevel Analysis. Multilevel Models Project.
Acknowledgments
London, UK: Institute of Education, University of London, 1995.
Funding was provided by the International Clinical Epi- [7] Bryk AS, Raudenbush SW. Hierarchical Linear Models: Application
and Data Analysis Methods. Newbury Park, CA, USA: Sage Produc-
demiology Network (INCLEN). tions, 1992.
[8] Heller R, O’Connell R, Lim L, Atallah A, Lanas F, Joshi P, Tatsan-
avivat P. Variation in management of acute myocardial infarction in 5
Appendix A countries. Int J Cardiol 1999;68:63–7.
[9] Sitthi-amorn C, Ngamvithayapongse J. The role of media and com-
The case history: “You see a previously healthy 60-year- munication in improving the use of drugs and other technologies. Int
old patient who was admitted to hospital one week earlier J Technology Assessment in Health Care 1998;14:71–80.
with a fractured hip. She has had fevers, cough, and produc- [10] American Thoracic Society. Hospital-acquired pneumonia in adults:
tive sputum for two days. Clinical examination reveals a Diagnosis, assessment of severity, initial antimicrobial therapy, and
temperature of 38.58C (1018F), her sputum is green, and her preventative strategies. A consensus statement. Am J Respir Crit Care
Med 1995;153:1711–25.
respiratory rate is mildly elevated. There is some dullness at [11] WHO. Acute respiratory infections in children: case management in
the base of the right lung field with breath sounds that are small hospitals in developing countries. A manual for doctors and
marginally louder than the opposite side.” other senior health workers. Geneva, Switzerland: World Health Or-
ganisation, 1990.
[12] Cunah BA. The antibiotic treatment of community-acquired, atypical,
and nosocomial pneumonias. Med Clin NA 1995;79:581–7.
References [13] Clarke CW. MJA Practice Essentials 5. Lower respiratory tract infec-
tion. Med J Aust 1997;167:42–7.
[1] Sackett DL, Rosenberg WMC, Gray JAM, Haynes RB. Evidence [14] Jamison DT, Mosley WH. Disease control priorities in developing
based medicine: What it is and what it isn’t. Br Med J 1996;312:71–2. countries: health policy responses to epidemiological change. Am J
[2] Ellis J, Mulligan I, Rowe J, Sackett DL. Inpatient general medicine is Public Health 1991;81:15–22.