Conclusions. General practitioners need to re-examine their own prescribing for URTI and decide whether it is consistent with current guidelines. Rational prescribing is not just part of the professional role of doctors, but will go a long way to impede the emergence of antibiotic resistance.

Keywords: Respiratory tract infections, antibiotics, physician practice patterns, family practice

Introduction
Upper respiratory tract infection is a common problem seen in primary care. Although it is generally believed that there is over-prescription of antibiotic for URTI in general practice, there are few reports of this in the indexed literature from the Asia-Pacific region.⁽¹⁾ Chang et al.⁽¹⁾ in a one-week prescription study in Taiwan, reported that antibiotic was prescribed in 31.3% of patients with common cold. Aljunid ⁽²⁾ reported a marked difference in the antibiotic prescription rates for URTI in private and government clinics in one district (75.9% and 45.5% respectively).

While antibiotic is credited with the dramatic reduction in the morbidity and mortality associated with many bacterial infections, its abuse has resulted in the rapid emergence of resistant strains that reduce the effectiveness of many antibiotics.⁽³⁾ It has been recognized that the problem of antibiotic abuse is a result of a complex interplay of various socio-cultural, economic and cognitive factors at the level of the patients, the prescribers and the drug industry.⁽⁴⁾

In this study, we report the factors associated with the prescribing of antibiotic for URTI in Malaysian general practice. The data were extracted from the study "Cost and Quality of Care in Three Urban Areas in Malaysia". ^(5,6)

Materials and Methods

Sampling and setting
A list of private general practice clinics (n=639) was created by perusal of the list of registered medical practitioners from Malaysian Medical Council (1995). Proportionate random samples of 150 clinics were selected (Kuala Lumpur 101, Penang 25 and Ipoh 24). These clinics were approached by mail and later visited personally by research assistants. The study was conducted over six months in the year 1999.

Questionnaire and definitions
In the morbidity component of this study, participating general practitioners were requested to complete Data Encounter Form for each of the 30 consecutive adult patients seen (age ³18 years). The Data Encounter Form was a 2-page questionnaire that asked for the following information from the clinical encounter: demographic data, reasons for encounter (RFEs, up to 5), physical findings, diagnoses (up to 2), investigation ordered, outpatient procedures performed, medical certificate given, medication prescribed (up to 8, but only a maximum of 5 items were analysed), and referral made.

The sources of payment were originally coded as cash (out-of-pocket payment by patients), panel (patients or clinics claiming the consultation fee from the employers) and managed care (payment is paid via a managed care organisation, primarily by capitation). Panel system and managed care were recoded as "non-cash" as subsequent analyses showed that they are similar for patient's demographic characteristics, morbidity and process of care.

Data analysis
We used SPSS version 10 for data entry and analysis. The morbidity data was coded using ICPC-2, ⁽⁷⁾ this was facilitated by using ICPC-2 plus Demonstrator (courtesy of Family Medicine Research Unit, Dept of General Practice, University of Sydney).

Categorical and continuous variables were compared using c2-test and t-test respectively. Statistical significance is set at p<0.05. Multinomial logistic regression was used to look for independent association between age, Chinese ethnicity, fever and sore throat and antibiotic prescription.

Results
One hundred and twenty five clinics returned 3481 complete Data Encounter Forms. Encounters with ICPC Diagnosis codes R74, R75 and R76 were classified as URTI (Table 1). We have classified as URTI in 11 encounters where diagnoses were recorded as symptoms only (R21 and R25) after verifying the chief complaints recorded were consistent with URTI. Thus the number (%) of encounter was 940 (27.0%).

Table 1: ICPC diagnosis codes of URTI encounters

Demographic characteristics
As shown in Table 2, consultations for upper respiratory tract infections were more common in males, those in the non-cash payment system and those who were employed. URTI was less commonly seen in the 'Other' ethnic group who were mostly immigrant workers.

Table 2: Demographic characteristics of patients with and without upper respiratory tract infections

Factors influencing antibiotic prescribing
Antibiotic was prescribed for 33.4% of all encounters and for 68.4% (95%CI 65.4%, 71.4%) diagnosed to be URTI. Antibiotic was prescribed in 67.2% of encounters coded as R74 (acute upper respiratory infection) and 90.7% of encounters coded as R76 (acute tonsillitis). Antibiotic was prescribed more commonly in patients with URTI than those without URTI (68.4% versus 20.5%, c2=707.36, p<0.001). URTI contributed 55.2% of all antibiotics prescribed in his study.

Antibiotic prescribing did not differ by gender (males 70.4%, females 66.2%, c²=1.902, p=0.168), payment system (cash 70.1%, non-cash 67.4%, c²=0.78, p=0.377) and employment status (employed 67.9%, unemployed 71.6%, c²=0.757, p=0.384).

Antibiotic prescribing differed significantly by ethnic groups (Malays 61.5%, Chinese 77.1%, Indians 76.9%, Others 69.7%; c²=24.76, p<0.001). Patients prescribed antibiotic were slightly older than those not given antibiotic (mean ages 34.2 versus 32.2 years, t=2.65, p=0.008).

Antibiotic prescribing for various URT symptoms were as follow: fever 72.7%, cough 68.9%, sore throat 78.8%, runny nose 65.5%, phlegm 66.7% and hoarseness 66.7%. Fever (c²=6.596, p=0.01) and sore throat (c²=27.467, p<0.001) were significantly associated with antibiotic prescribing.

Four factors that were initially associated with antibiotic prescribing (age, Chinese ethnicity, fever, sore throat) were tested with multinomial logistic regression; the last three factors remained independently associated with antibiotic prescribing (Table 3).

Table 3: Unadjusted and adjusted odds ratios of factors associated with antibiotic prescribing

Antibiotic choice
Table 4 showed the antibiotic group and six most common antibiotics prescribed in the patients with URTI.

Table 4: Antibiotic choice in URTI

* aminoglycosides, quinolones, metronidazole, lincomycin
**in 13 encounters two antibiotics were prescribed.

Discussion
The research methodology in this study relies on the conscientious recording by the participating general practitioners, for which we were unable to verify by direct examination of the actual clinical records. Nonetheless, we feel that the data collected broadly apply to the adult attendees in the urban Malaysian general practice for the following reasons: (a) high response rate (83.3% of clinics approached returned the encounter forms), (b) the profile of the general practitioners in this study was similar to those in the West Malaysia,^(6,8) (c) missing data for medication in URTI encounters was only 1.7% (data not shown), and (d) similarly high antibiotic prescribing rate for URTI had been reported in a smaller Malaysian general practice survey.⁽²⁾

URTI contributed at least one-quarter of all encounters sampled from the general practice clinics. The differences in the consultation rates for URTI for various demographic subgroups suggest the complex interplay of many factors that lead to a consultation for URTI in general practice.

Antibiotic was prescribed for 68.4% of the encounters diagnosed to have URTI. The prescribing rate was high irrespectively of the demographic characteristics and clinical features, although we found that Chinese ethnicity, the history of fever and sore throat were significantly associated with antibiotic prescription. The poor recording of physical findings in this morbidity survey did not allow a more detailed analysis of the clinical decision-making of the general practitioners. However, the prescribing of antibiotic is somewhat indiscriminate and appears to be inconsistent with the guidelines.^(9,10)

The choice of antibiotics was by and large appropriate with respect to their ability to eradicate the group A b-haemolytic streptococcus, the most important bacterial pathogen in URTI. However, some choices were clearly inappropriate on account of the sensitivity towards group A b-haemolytic streptococcus (e.g. sulphonamides, tetracyclines, metronidazole).^(9,10) The prescribing of newer antibiotics (e.g. azithromycin and quinolones) was relatively uncommon.

This study illustrates that antibiotic prescribing for URTI in Malaysian general practice is a suitable target of intervention since it contributed at least half of all antibiotics prescribed in general practice. Snow et al, ⁽⁹⁾ in keeping with systematic review⁽¹¹⁾ of the predictive features of streptococcal pharyngitis, recommended starting empirical antibiotics (either penicillin V, amoxycillin or erythromycin) for adults with at least three of four clinical criteria (history of fever, tonsillar exudate, tender anterior cervical lymphadenopathy, and absence of cough). However, the problem of patient expectation, either implied or expressed, for antibiotic in general practice is keenly felt by general practitioners and possibly influence their prescribing behaviour.⁽¹²⁾ Thus clinical trials aimed at increasing adherence to evidence-based guidelines in URTI have had mixed success so far. ^(13-15)

Acknowledgement
We wish to thank the Family Medicine Research Unit, Department of General Practice, University of Sydney for providing a copy of ICPC-2 Demonstrator. We greatly appreciated the willingness of the participating general practitioners for providing the clinical data. This project was funded by the Intensified Research in Priority Areas (IRPA Grant, Code no: 06-02-02-0061).

References
1. Chang SC, Shiu MN, Chen TJ. Antibiotic usage in primary care units in Taiwan after the institution of national health insurance. Diagn Microbiol Infect Dis 2001 Jul; 40(3): 137-43

2. Aljunid S. Management of upper respiratory tract infections by public and private sector doctors in a rural district of Malaysia. Paper presented in Annual Scientific Meeting, Academy of Medicine, Malaysia on 24th March, 1996.

3. Lim VKE. Antibiotic resistance and its control in the Far East. Antibiotics Chemotherapy (Newsletter of the International Society of Chemotherapy) 2001 September; 5(2): 1-3

4. Okeke IN, Lamikanra A, Iman RE. Socioeconomic and behavioral factors leading to acquired bacterial resistance to antibiotics in developing countries. Emerging Infectious Diseases 1999; 5(1): 18-27

5. Aljunid SM, Cheah M, Soe Nyunt-U, Kwa SK, Rohaizat Yon, Ding LM. Cost analysis of private primary care services in three urban centres in Malaysia. Malaysian Journal of Public Health Medicine 2000; 1: 8-15

6. Teng CL, Aljunid SM, Cheah M, Leong KC, Kwa SK. Morbidity and process of care in urban Malaysian general practice: the impact of payment system. Med J Malaysia 2003; 58(3): 365-374

7. Classification Committee of World Organization of Family Doctors. ICPC-2: International Classification of Primary Care. Oxford: Oxford University Press, 1997.

8. Khoo EM, Tan PL. Profile of general practices in Malaysia. Asia Pac J Public Health 1998; 10: 81-87

9. Snow V, Mottur-Pilson C, Cooper RJ, Hoffman JR. Principles of appropriate antibiotic use for acute pharyngitis in adults. Ann Intern Med 2001; 134: 506-508

10. Word Health Organization. WHO Model Prescribing Information: Drugs used in the treatment of streptococcal pharyngitis and prevention of rheumatic fever. WHO/EDM/PAR/99.1. Geneva: World Health Organization

11. Ebell MH, Smith MA, Barry HC, Ives K, Carey M. Does this patient have strep throat? JAMA 2000; 284(22): 2912-2918

12. Butler CC, Rollnick S, Pill R. Maggs-Rapport F, Stott N. Understanding the culture of prescribing: qualitative study of general practitioners' and patients' perceptions of antibiotics for sore throats. BMJ 1998; 317: 638-642

13. O'Connor PJ, Amundson G, Christianson J. Performance failure of an evidence-based upper respiratory infection clinical guideline. J Fam Pract 1999; 48(9): 690-697

14. Raz R, Porat V, Ephros M. Can an educational program improve the diagnosis and treatment of pharyngotonsillitis in the ambulatory care setting? Isr J Med Sci 1995; 31: 432-435

15. Zwar N, Wolk J, Gordon J, Sanson-Fisher R, Kehoe L. Influencing antibiotic prescribing in general practice: a trial of prescriber feedback and management guidelines. Fam Pract 1999; 16: 495-500