Predicting bacterial cause in infectious conjunctivitis: cohort study on informativeness of combinations of signs and symptoms
http://www.100md.com
《英国医生杂志》
1 Division of Clinical Methods and Public Health, Department of General Practice, Academic Medical Centre, University of Amsterdam, Meibergdreef 15, 1105 AZ, Amsterdam, Netherlands, 2 Horten-Zentrum, Zurich, Switzerland, 3 Medical Centre Alkmaar, Alkmaar, Netherlands
Correspondence to: R P Rietveld r.p.rietveld@amc.uva.nl
Abstract
In the developed world, acute infectious conjunctivitis is a common disorder with an annual incidence of 1.5-2% in primary care.1-5 Randomised trials in patients with suspected acute bacterial conjunctivitis show a pooled prevalence of bacterial pathogens of 50% (95% confidence interval 45% to 54%).6-9 No more than half of the cases of acute infectious conjunctivitis in primary care probably have a bacterial origin. Confronted with acute infectious conjunctivitis, most general practitioners feel unable to discriminate between a bacterial and a viral cause. In practice, more than 80% of such patients receive antibiotics.1 5 Hence, in cases of acute infectious conjunctivitis, many unnecessary ocular antibiotics are prescribed. In 2001 in the Netherlands, more than 900 000 prescriptions for topical ocular antibiotics were issued, at a cost of 8.85 million (￡5.9 million, $10.9 million). In England 3.4 million community prescriptions for these antibiotics are issued each year, at a cost to the NHS of ￡4.7 million (7.1 million, $8.7 million).10 11
To select those patients who might benefit most from antibiotic treatment, the general practitioner needs an informative diagnostic tool to determine a bacterial cause. With such a tool, antibiotic prescriptions may be reduced and better targeted. Most general practitioners make the distinction between a bacterial cause and another cause on the basis of signs and symptoms. Additional diagnostic investigations, such as a culture of the conjunctiva, are seldom done, mostly because of the resulting diagnostic delay. Can general practitioners actually differentiate between bacterial and viral conjunctivitis on the basis of signs and symptoms alone? Major textbooks list several signs and symptoms that are supposed to be diagnostic for the cause of acute infectious conjunctivitis.12-14 A recently published systematic literature search summed up the signs and symptoms and found no evidence for these assertions.15 This paper presents what seems to be the first empirical study on the diagnostic informativeness of signs and symptoms in acute infectious conjunctivitis.
Methods
Between September 1999 and December 2002 we enrolled 184 patients; data from 177 (96%) of these could be analysed (fig). The reasons for non-inclusion were refusal (n = 2) or incompleteness of data (n = 5). Three patients had incomplete index tests, and the culture results for two patients were unknown because the culture samples never arrived at the laboratory.
Flow of participants through the study
The prevalence of a positive bacterial culture in the study eye was 32% (57/177). The groups (culture positive and culture negative) were comparable with respect to baseline demographics, but some notable differences existed in the results of index tests (table 1). A history of conjunctivitis occurred more often in participants with a negative culture (21% v 9%). In the group with a positive culture, more patients had two glued eyes in the morning (39% v 11%) and bilateral involvement (37% v 16%). The most prevalent species was Streptococcus pneumoniae, which accounted for 27/57 of the positive cultures (table 2).
Table 1 Baseline demographic characteristics and index test results and their univariate odds ratios. Values are numbers (percentages) unless stated otherwise. The prevalence of a positive culture was 32% (57/177)
Table 2 Culture results
Three determinants were retained in the multivariable regression analysis: history of conjunctivitis (yes or no), itch (yes or no), and glued eyes in the morning (0, 1, or 2). Table 3 lists the odds ratios of these independent indicators of a positive bacterial culture and their clinical scores. We found no statistical interactions.
Table 3 Results of logistic regression analysis. Independent indicators of positive bacterial culture and their clinical score
The area under the receiver operating characteristics curve of the final model was 0.74 (95% confidence interval 0.65 to 0.82). The Hosmer-Lemeshow goodness of fit test had a P value of 0.117, indicating that the model does not misrepresent the data.17 Validation of this model with the bootstrap technique showed hardly any indication of undue influence by particular patients (corrected 95% confidence interval of area under curve 0.63 to 0.80).
The logistic regression analysis generated 12 different combinations of test results. These combinations corresponded to nine different clinical scores, varying from +5 to -3. For each clinical score, we calculated the probability of a positive culture. For a patient with a clinical score of +5, this probability was increased from 32% (prevalence in this study) to 77% (table 4). By contrast, a clinical score of -3 lowered this probability to 4%. Table 4 allows the calculation of the numbers of correctly treated patients (sensitivity) and correctly untreated patients (specificity) and the reduction of prescriptions with different treatment cut-off points. For example, if the treatment cut-off point is set at +2, indicating that only patients with a clinical score of +2 or higher receive ocular antibiotics, 38/57 (67%) of patients are correctly treated and 87/120 (73%) patients are correctly untreated. If applied to our study population, the cut-off point of +2 would lead to a reduction in prescriptions of antibiotics from more than 80% (current practice) to 40% (71/177).
Table 4 Clinical scores and their associated probabilities of a positive culture, sensitivity, and specificity
Discussion
Okkes IM, Oskam SK, Lamberts H. Van klacht naar diagnose: episodegegevens uit de huisartspraktijk. Bussum: Coutinho, 1998.
Van der Werf GT, Smit RJA, Stewart RE, Meyboom de Jong B. Spiegel op de huisarts: over registratie van ziekte, medicatie en verwijzing in de geautomatiseerde huisartspraktijk. Groningen: Rijksuniversiteit Groningen, 1998.
Van de Lisdonk EH, Bakx J. Continue morbiditeits registratie: ziekten in de huisartspraktijk. Bunge: Elsevier, 1999.
Sheikh A, Hurwitz B. Topical antibiotics for acute bacterial conjunctivitis: a systematic review. Br J Gen Pract 2001;51: 473-7.
Everitt H, Little P. How do GPs diagnose and manage acute infective conjunctivitis? A GP survey. Fam Pract 2002;19: 658-60.
Horven I. Acute conjunctivitis: a comparison of fusidic acid viscous eye drops and chloramphenicol. Acta Ophthalmol 1993;71: 165-8.
Miller IM, Wittreich J, Vogel R, Cook TJ. The safety and efficacy of topical norfloxacin compared with placebo in the treatment of acute, bacterial conjunctivitis. Eur J Ophthalmol 1992;2: 58-66.
Gallenga PE, Lobefalo L, Colangelo L, Della Loggia G, Orzalesi N, Velati P, et al. Topical lomefloxacin 0.3% twice daily versus tobramycin 0.3% in acute bacterial conjunctivitis: a multicenter double-blind phase III study. Ophthalmologica 1999;213: 250-7.
Agius-Fenandez A, Patterson A, Fsadni M, Jauch A, Raj PS. Topical lomefloxacin versus topical chloramphenicol in the treatment of acute bacterial conjunctivitis. Clin Drug Invest 1998;15: 263-9.
Genees en hulpmiddelen Informatie Project. Annual report prescription data. College voor zorgverzekeringen, Amstelveen, 2001.
Department of Health. Prescription cost analysis data. Leeds: Department of Health, 1998.
Krachmer JH. Cornea. St Louis: Mosby, 1997.
Kanski JJ. Clinical ophthalmology: a systematic approach. Oxford: Butterworth-Heinemann, 1999.
Tasman W, Jaeger EA. Duane's clinical ophthalmology on CD-Rom. Lippincot Williams and Wilkins, 2001.
Rietveld RP, Van Weert HC, Riet ter G, Bindels PJ. Diagnostic impact of signs and symptoms in acute infectious conjunctivitis: systematic literature search. BMJ 2003;327: 789.
Hall GS, Pezzlo M. Ocular cultures. In: Isenberg HD, ed. Clinical microbiology procedures handbook. Washington: American Society for Microbiology, 1995.
Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, 1989.
Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15: 361-87.
Brown EM, Thomas P. Fusidic acid resistance in Staphylococcus aureus isolates. Lancet 2002;359: 803.
Mills O Jr, Thornsberry C, Cardin CW, Smiles KA, Leyden JJ. Bacterial resistance and therapeutic outcome following three months of topical acne therapy with 2% erythromycin gel versus its vehicle. Acta Derm Venereol 2002;82: 260-5.
Goldstein MH, Kowalski RP, Gordon YJ. Emerging fluoroquinolone resistance in bacterial keratitis: a 5-year review. Ophthalmology 1999;106: 1313-8.(Remco P Rietveld, general)
Correspondence to: R P Rietveld r.p.rietveld@amc.uva.nl
Abstract
In the developed world, acute infectious conjunctivitis is a common disorder with an annual incidence of 1.5-2% in primary care.1-5 Randomised trials in patients with suspected acute bacterial conjunctivitis show a pooled prevalence of bacterial pathogens of 50% (95% confidence interval 45% to 54%).6-9 No more than half of the cases of acute infectious conjunctivitis in primary care probably have a bacterial origin. Confronted with acute infectious conjunctivitis, most general practitioners feel unable to discriminate between a bacterial and a viral cause. In practice, more than 80% of such patients receive antibiotics.1 5 Hence, in cases of acute infectious conjunctivitis, many unnecessary ocular antibiotics are prescribed. In 2001 in the Netherlands, more than 900 000 prescriptions for topical ocular antibiotics were issued, at a cost of 8.85 million (￡5.9 million, $10.9 million). In England 3.4 million community prescriptions for these antibiotics are issued each year, at a cost to the NHS of ￡4.7 million (7.1 million, $8.7 million).10 11
To select those patients who might benefit most from antibiotic treatment, the general practitioner needs an informative diagnostic tool to determine a bacterial cause. With such a tool, antibiotic prescriptions may be reduced and better targeted. Most general practitioners make the distinction between a bacterial cause and another cause on the basis of signs and symptoms. Additional diagnostic investigations, such as a culture of the conjunctiva, are seldom done, mostly because of the resulting diagnostic delay. Can general practitioners actually differentiate between bacterial and viral conjunctivitis on the basis of signs and symptoms alone? Major textbooks list several signs and symptoms that are supposed to be diagnostic for the cause of acute infectious conjunctivitis.12-14 A recently published systematic literature search summed up the signs and symptoms and found no evidence for these assertions.15 This paper presents what seems to be the first empirical study on the diagnostic informativeness of signs and symptoms in acute infectious conjunctivitis.
Methods
Between September 1999 and December 2002 we enrolled 184 patients; data from 177 (96%) of these could be analysed (fig). The reasons for non-inclusion were refusal (n = 2) or incompleteness of data (n = 5). Three patients had incomplete index tests, and the culture results for two patients were unknown because the culture samples never arrived at the laboratory.
Flow of participants through the study
The prevalence of a positive bacterial culture in the study eye was 32% (57/177). The groups (culture positive and culture negative) were comparable with respect to baseline demographics, but some notable differences existed in the results of index tests (table 1). A history of conjunctivitis occurred more often in participants with a negative culture (21% v 9%). In the group with a positive culture, more patients had two glued eyes in the morning (39% v 11%) and bilateral involvement (37% v 16%). The most prevalent species was Streptococcus pneumoniae, which accounted for 27/57 of the positive cultures (table 2).
Table 1 Baseline demographic characteristics and index test results and their univariate odds ratios. Values are numbers (percentages) unless stated otherwise. The prevalence of a positive culture was 32% (57/177)
Table 2 Culture results
Three determinants were retained in the multivariable regression analysis: history of conjunctivitis (yes or no), itch (yes or no), and glued eyes in the morning (0, 1, or 2). Table 3 lists the odds ratios of these independent indicators of a positive bacterial culture and their clinical scores. We found no statistical interactions.
Table 3 Results of logistic regression analysis. Independent indicators of positive bacterial culture and their clinical score
The area under the receiver operating characteristics curve of the final model was 0.74 (95% confidence interval 0.65 to 0.82). The Hosmer-Lemeshow goodness of fit test had a P value of 0.117, indicating that the model does not misrepresent the data.17 Validation of this model with the bootstrap technique showed hardly any indication of undue influence by particular patients (corrected 95% confidence interval of area under curve 0.63 to 0.80).
The logistic regression analysis generated 12 different combinations of test results. These combinations corresponded to nine different clinical scores, varying from +5 to -3. For each clinical score, we calculated the probability of a positive culture. For a patient with a clinical score of +5, this probability was increased from 32% (prevalence in this study) to 77% (table 4). By contrast, a clinical score of -3 lowered this probability to 4%. Table 4 allows the calculation of the numbers of correctly treated patients (sensitivity) and correctly untreated patients (specificity) and the reduction of prescriptions with different treatment cut-off points. For example, if the treatment cut-off point is set at +2, indicating that only patients with a clinical score of +2 or higher receive ocular antibiotics, 38/57 (67%) of patients are correctly treated and 87/120 (73%) patients are correctly untreated. If applied to our study population, the cut-off point of +2 would lead to a reduction in prescriptions of antibiotics from more than 80% (current practice) to 40% (71/177).
Table 4 Clinical scores and their associated probabilities of a positive culture, sensitivity, and specificity
Discussion
Okkes IM, Oskam SK, Lamberts H. Van klacht naar diagnose: episodegegevens uit de huisartspraktijk. Bussum: Coutinho, 1998.
Van der Werf GT, Smit RJA, Stewart RE, Meyboom de Jong B. Spiegel op de huisarts: over registratie van ziekte, medicatie en verwijzing in de geautomatiseerde huisartspraktijk. Groningen: Rijksuniversiteit Groningen, 1998.
Van de Lisdonk EH, Bakx J. Continue morbiditeits registratie: ziekten in de huisartspraktijk. Bunge: Elsevier, 1999.
Sheikh A, Hurwitz B. Topical antibiotics for acute bacterial conjunctivitis: a systematic review. Br J Gen Pract 2001;51: 473-7.
Everitt H, Little P. How do GPs diagnose and manage acute infective conjunctivitis? A GP survey. Fam Pract 2002;19: 658-60.
Horven I. Acute conjunctivitis: a comparison of fusidic acid viscous eye drops and chloramphenicol. Acta Ophthalmol 1993;71: 165-8.
Miller IM, Wittreich J, Vogel R, Cook TJ. The safety and efficacy of topical norfloxacin compared with placebo in the treatment of acute, bacterial conjunctivitis. Eur J Ophthalmol 1992;2: 58-66.
Gallenga PE, Lobefalo L, Colangelo L, Della Loggia G, Orzalesi N, Velati P, et al. Topical lomefloxacin 0.3% twice daily versus tobramycin 0.3% in acute bacterial conjunctivitis: a multicenter double-blind phase III study. Ophthalmologica 1999;213: 250-7.
Agius-Fenandez A, Patterson A, Fsadni M, Jauch A, Raj PS. Topical lomefloxacin versus topical chloramphenicol in the treatment of acute bacterial conjunctivitis. Clin Drug Invest 1998;15: 263-9.
Genees en hulpmiddelen Informatie Project. Annual report prescription data. College voor zorgverzekeringen, Amstelveen, 2001.
Department of Health. Prescription cost analysis data. Leeds: Department of Health, 1998.
Krachmer JH. Cornea. St Louis: Mosby, 1997.
Kanski JJ. Clinical ophthalmology: a systematic approach. Oxford: Butterworth-Heinemann, 1999.
Tasman W, Jaeger EA. Duane's clinical ophthalmology on CD-Rom. Lippincot Williams and Wilkins, 2001.
Rietveld RP, Van Weert HC, Riet ter G, Bindels PJ. Diagnostic impact of signs and symptoms in acute infectious conjunctivitis: systematic literature search. BMJ 2003;327: 789.
Hall GS, Pezzlo M. Ocular cultures. In: Isenberg HD, ed. Clinical microbiology procedures handbook. Washington: American Society for Microbiology, 1995.
Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, 1989.
Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15: 361-87.
Brown EM, Thomas P. Fusidic acid resistance in Staphylococcus aureus isolates. Lancet 2002;359: 803.
Mills O Jr, Thornsberry C, Cardin CW, Smiles KA, Leyden JJ. Bacterial resistance and therapeutic outcome following three months of topical acne therapy with 2% erythromycin gel versus its vehicle. Acta Derm Venereol 2002;82: 260-5.
Goldstein MH, Kowalski RP, Gordon YJ. Emerging fluoroquinolone resistance in bacterial keratitis: a 5-year review. Ophthalmology 1999;106: 1313-8.(Remco P Rietveld, general)