A protocol improves GP recording of long-term sickness absence risk factors

BACKGROUND

If general practitioners (GPs) were better informed about patients' risks of long-term sickness absence (LTSA), they could incorporate these risk assessments into their patient management plans and cooperate more with occupational physicians to prevent LTSA.

AIM

To evaluate the effectiveness of a protocol helping GPs in recording risks of LTSA and in co-operating with occupational physicians (OPs).

METHODS

Twenty-six GPs (co-operating in four groups) in Amsterdam, The Netherlands, participated in a controlled intervention study. Fourteen GPs were the protocol-supported intervention group and twelve GPs were the reference group. Outcome measures were consultations containing work-related information, information about two risk factors for LTSA, referrals to OPs and contacts of OPs with GPs and patients. Outcomes were identified through an electronic search in the GPs' information systems. Entries containing information were independently scored by two investigators. The proportions of patients with consultations documenting LTSA-pertinent items were compared between the groups, accounting for differences at baseline.

RESULTS

There was no increase in consultations containing work-related information. Recording of risk factor information increased in the intervention group; the difference was 4.5% [95% CI 1.5-7.6] and 1.8% (95% CI -0.8 to 4.4) for the two risk factors. The referral rate to the OP increased by 2.9% (95% CI 1.2-4.5). There was no effect on contacts of OPs with GPs or with patients.

CONCLUSION

Protocol-supported consultations may lead to a modest increase in information regarding two risk factors for LTSA in GPs' electronic records and to more referrals to OPs.

Use of uterine artery Doppler ultrasonography to predict pre-eclampsia and intrauterine growth restriction: a systematic review and bivariable meta-analysis

BACKGROUND

Alterations in waveforms in the uterine artery are associated with the development of pre-eclampsia and intrauterine growth restriction. We investigated the predictive accuracy of all uterine artery Doppler indices for both conditions in the first and second trimesters.

METHODS

We identified relevant studies through searches of MEDLINE, EMBASE, the Cochrane Library and Medion databases (all records to April 2006) and by checking bibliographies of identified studies and consulting with experts. Four of us independently selected studies, extracted data and assessed study validity. We performed a bivariable meta-analysis of sensitivity and specificity and calculated likelihood ratios.

RESULTS

We identified 74 studies of pre-eclampsia (total 79,547 patients) and 61 studies of intrauterine growth restriction (total 41 131 patients). Uterine artery Doppler ultrasonography provided a more accurate prediction when performed in the second trimester than in the first-trimester. Most Doppler indices had poor predictive characteristics, but this varied with patient risk and outcome severity. An increased pulsatility index with notching was the best predictor of pre-eclampsia (positive likelihood ratio 21.0 among high-risk patients and 7.5 among low-risk patients). It was also the best predictor of overall (positive likelihood ratio 9.1) and severe (positive likelihood ratio 14.6) intrauterine growth restriction among low-risk patients.

INTERPRETATION

Abnormal uterine artery waveforms are a better predictor of pre-eclampsia than of intrauterine growth restriction. A pulsatility index, alone or combined with notching, is the most predictive Doppler index. These indices should be used in clinical practice. Future research should also concentrate on combining uterine artery Doppler ultrasonography with other tests.

Self-referrals to the A&E department during out-of-hours: patients' motives and characteristics

OBJECTIVE

To determine self-referrals' motives to visit the accident & emergency departments (AED) and to compare their characteristics to patients contacting the GP cooperative.

METHODS

Postal questionnaires were send to AED self-referrals and logistic regression analysis was used to contrast self-referrals to patients contacting the GP cooperative.

RESULTS

For a study population of 62,000, during 4 months, 5547 contacts were registered with the GP cooperative, along with 808 AED contacts, 344 of whom (43%) were self-referrals. Main reasons to visit the AED were the perceived need for diagnostic facilities and the conviction that the hospital specialist was best qualified to handle the problem. Dissatisfaction with the GP cooperative among respondents was high. Self-referral to the AED was positively associated with injury, age between 15 and 64, musculoskeletal, cardiovascular and respiratory problems, and distance to the GP centre.

CONCLUSION

Self-referrals emerge as patients with a strong preference for the AED, mainly based on assumptions on quality of care and necessary facilities.

PRACTICE IMPLICATIONS

While self-referrals may, in part, make motivated and appropriate choices to visit the AED, new integrated care models should be studied that can adequately deflect those who are eligible for GP care.

Demographic characteristics and quality of life of patients with unexplained complaints: a descriptive study in general practice

Objective

About 13% of GPs’ consultations involve unexplained complaints (UCs). These complaints can progress to chronic conditions like medically unexplained symptoms, chronic functional symptoms or somatoform disorders. Little is known about the demographic characteristics and quality of life of patients with early stage UCs. Our study objective was to describe these characteristics. Additionally we compared them with other patient groups to serve as a frame of reference.

Methods

Descriptive study in general practices. Patients with early stage UCs who had not had elaborate diagnostic investigations were included. Demographic characteristics were compared to a Dutch general practice population. Quality of life scores were measured with the RAND-36 and compared to another Dutch general practice population and to depressed patients.

Results

Data of 466 patients were available for analysis. Mean age was 44 years and 74% were females, mostly higher educated. Of the patients, 63% presented with unexplained fatigue. On average, quality of life was poor (mean RAND-36 domain scores 37–73), also in comparison with other groups.

Conclusion

General practice patients presenting with UCs have a remarkably poor quality of life. Future research should explore how early identification of patients at risk of developing chronicity can take place. Awareness of potential poor quality of life may influence GPs' medical decision making.

Do general practitioners adhere to the guideline on infectious conjunctivitis? Results of the Second Dutch National Survey of General Practice

Background

In 1996 the guideline 'The Red Eye' was first published by the Dutch College of General Practitioners. The extent to which general practitioners adhere to this guideline is unclear. Recently, data on the management of infectious conjunctivitis by general practitioners became available from the Second Dutch National Survey of General Practice. We measured the age-specific incidence of infectious conjunctivitis, described its management by Dutch general practitioners, and then compared these findings with the recommendations made in the guideline.

Methods

In 2001, over a 12-month period, data from all patient contacts with 195 general practitioners were taken from electronic medical records. Registration was episode-oriented; all consultations dealing with the same health problem were grouped into disease episodes. Data concerning all episodes of infectious conjunctivitis (ICPC-code F70 and sub codes) were analysed.

Results

Over one year, 5,213 new and recurrent episodes of infectious conjunctivitis were presented to general practitioners from a population of N = 375,899, resulting in an overall incidence rate of 13.9 per 1000 person-years, varying from more than 80/1000 py in children up to one-year old, to less than 12/1000 py in children over the age of 4. Topical ophthalmic ointments were prescribed in 87% of the episodes, of which 80% was antibiotic treatment. Fusidic acid gel was most frequently prescribed (69%). In most episodes general practitioners did not adhere to the guideline.

Conclusion

In 2001, the management of infectious conjunctivitis by Dutch general practitioners was not in accordance with the recommendations of the consensus-based guideline published five years previously, despite its wide distribution. In 2006 this guideline was revised. Its successful implementation requires more than distribution alone. Probably the most effective way to achieve this is by following a model for systemic implementation.

Nurse telephone triage in out-of-hours GP practice: determinants of independent advice and return consultation

BACKGROUND

Nowadays, nurses play a central role in telephone triage in Dutch out-of-hours primary care. The percentage of calls that is handled through nurse telephone advice alone (NTAA) appears to vary substantially between GP cooperatives. This study aims to explore which determinants are associated with NTAA and with subsequent return consultations to the GP.

METHODS

For the ten most frequently presented problems, a two-week follow-up cohort study took place in one cooperative run by 25 GPs and 8 nurses, serving a population of 62,291 people. Random effects logistic regression analysis was used to study the determinants of NTAA and return consultation rates. The effect of NTAA on hospital referral rates was also studied as a proxy for severity of illness.

RESULTS

The mean NTAA rate was 27.5%--ranging from 15.5% to 39.4% for the eight nurses. It was higher during the night (RR 1.63, CI 1.48-1.76) and lower with increasing age (RR 0.96, CI 0.93-0.99, per ten years) or when the patient presented >2 problems (RR 0.65; CI 0.51-0.83). Using cough as reference category, NTAA was highest for earache (RR 1.49; CI 1.18-1.78) and lowest for chest pain (RR 0.18; CI 0.06-0.47). After correction for differences in case mix, significant variation in NTAA between nurses remained (p < 0.001). Return consultations after NTAA were higher after nightly calls (RR 1.23; CI 1.04-1.40). During first return consultations, the hospital referral rate after NTAA was 1.5% versus 3.8% for non-NTAA (difference -2.2%; CI -4.0 to -0.5).

CONCLUSION

Important inter-nurse variability may indicate differences in perception on tasks and/or differences in skill to handle telephone calls alone. Future research should focus more on modifiable determinants of NTAA rates.

Prediction of pre-eclampsia: a protocol for systematic reviews of test accuracy

BACKGROUND

Pre-eclampsia, a syndrome of hypertension and proteinuria, is a major cause of maternal and perinatal morbidity and mortality. Accurate prediction of pre-eclampsia is important, since high risk women could benefit from intensive monitoring and preventive treatment. However, decision making is currently hampered due to lack of precise and up to date comprehensive evidence summaries on estimates of risk of developing pre-eclampsia.

METHODS/DESIGN

A series of systematic reviews and meta-analyses will be undertaken to determine, among women in early pregnancy, the accuracy of various tests (history, examinations and investigations) for predicting pre-eclampsia. We will search Medline, Embase, Cochrane Library, MEDION, citation lists of review articles and eligible primary articles and will contact experts in the field. Reviewers working independently will select studies, extract data, and assess study validity according to established criteria. Language restrictions will not be applied. Bivariate meta-analysis of sensitivity and specificity will be considered for tests whose studies allow generation of 2 x 2 tables.

DISCUSSION

The results of the test accuracy reviews will be integrated with results of effectiveness reviews of preventive interventions to assess the impact of test-intervention combinations for prevention of pre-eclampsia.

More medical journals should inform their contributors about three key principles of graph construction

OBJECTIVES

To identify key principles of graph construction and to review guidance on graph construction provided by medical journals.

STUDY DESIGN AND SETTING

Review of a prevailing model of graph comprehension and survey of authors' instructions for all 120 core medical journals as indexed by PubMed.

RESULTS

Graph construction should meet three requirements for optimal graph comprehension: Data symbols should be clearly distinguishable to enable their visual detection. Second, the data should be arranged in such a way that estimation of values and important relationships in the data is facilitated, and third, there should be a full explanation of all graph elements and of how the data were obtained. Seven journals (5.8%) provided recommendations on graph construction. Six journals put emphasis primarily on the explanation of graph elements and, to a lesser extent, on the visual detection of data symbols. Recommendations of the seventh journal focused mostly on how to enhance estimation of important relationships. All journals presented single recommendations whose selection seemed unstructured and not based on key principles of graph construction.

CONCLUSION

Medical journals should order their instructions according to the three key aspects of graph construction.

Randomized consent designs in randomized controlled trials: Systematic literature search

BACKGROUND

Three types of randomized consent designs are distinguished and ranked according to the extent to which participants are informed about treatment options: single-consent (those in the experimental group learn about their assigned treatment), incomplete-double-consent (all participants learn about their assigned treatment), and complete-double-consent (all participants learn about all treatments studied). All are methodologically, ethically, and judicially controversial. Even so, their use is justified if blinding is deemed necessary, but impossible to achieve by sham procedures (placebo), and experimental treatment seems attractive to potential participants.

OBJECTIVE

The aim of this study is to give a comprehensive overview of the use of randomized consent designs. Data sources are MEDLINE (1/1977-2/2003), EMBASE (1/1984-2/2003), PsycINFO (1/1996-2/2003), the Cochrane Library, and the Science Citation Index database.

REVIEW METHODS

Eligible were studies using a randomized consent design. Cluster randomized trials were excluded. One reviewer selected and data-extracted eligible papers. A second reviewer independently data-extracted 10% of the papers. Data on country of study conduct, year of commencement, area of medicine, type of design, reason(s) for use, details on approval by a research ethics committee, the index and reference intervention, nature of endpoints, and details on collection of data were extracted. Furthermore, for each trial, the rates of non-compliance and loss to follow-up were registered by treatment arm. The three types of randomized consent designs were compared as to differences between the rates of non-compliance and loss to follow-up in the separate trial arms.

RESULTS

Randomized consent designs are seldom used (n=50). When used, they have often been used in the wrong circumstances (misuse). In 65% of the studies the non-compliance in the index group is larger than in the reference group. Contrary to expectation, trials using the incomplete-double design were associated with significantly higher rates of non-compliance and loss to follow-up in the reference groups than trials employing the other two versions.

CONCLUSION

Trialists and physicians should be aware of the proper indication for the use of randomized consent designs.

Sample sizes of studies on diagnostic accuracy: literature survey

OBJECTIVES

To determine sample sizes in studies on diagnostic accuracy and the proportion of studies that report calculations of sample size.

DESIGN

Literature survey.

DATA SOURCES

All issues of eight leading journals published in 2002.

METHODS

Sample sizes, number of subgroup analyses, and how often studies reported calculations of sample size were extracted.

RESULTS

43 of 8999 articles were non-screening studies on diagnostic accuracy. The median sample size was 118 (interquartile range 71-350) and the median prevalence of the target condition was 43% (27-61%). The median number of patients with the target condition--needed to calculate a test's sensitivity--was 49 (28-91). The median number of patients without the target condition--needed to determine a test's specificity--was 76 (27-209). Two of the 43 studies (5%) reported a priori calculations of sample size. Twenty articles (47%) reported results for patient subgroups. The number of subgroups ranged from two to 19 (median four). No studies reported that sample size was calculated on the basis of preplanned analyses of subgroups.

CONCLUSION

Few studies on diagnostic accuracy report considerations of sample size. The number of participants in most studies on diagnostic accuracy is probably too small to analyse variability of measures of accuracy across patient subgroups.

Accuracy of serum uric acid determination in predicting pre-eclampsia: A systematic review3-33

BACKGROUND

Pre-eclampsia is a leading cause of maternal and perinatal mortality and morbidity. Early prediction of pre-eclampsia may be used to target high-risk women for effective preventative treatments. Hyperuricemia is associated with pre-eclampsia and it has been tested in early pregnancy for its ability to predict the later onset of the disease.

OBJECTIVE

To systematically review the accuracy of determining serum uric acid before the 25th gestational week in predicting pre-eclampsia.

METHODS

We searched Medline, Embase, Cochrane Library, Medion, bibliographies of review articles and eligible primary studies, and contacted experts in the field. Language restrictions were not applied. Reviewers working independently selected studies, extracted data and assessed validity according to eight criteria. Sensitivities and specificities were plotted in a Receiver Operating Characteristic plot.

RESULTS

We identified five primary articles, with a total of 572 women, including 44 who developed pre-eclampsia. Study populations differed clinically, with incidences ranging from 3.4% to 40.1%. Sensitivity ranged from 0.0% to 55.6% and specificity ranged from 76.9% to 94.9%. Clinical heterogeneity and poor reporting precluded sensible pooling of data. In particular, consecutive entry of eligible women, blind assessment of the reference standard, and treatment to prevent pre-eclampsia were poorly reported.

CONCLUSION

Contrary to previous reviews, we conclude that there is currently insufficient evidence to draw firm conclusions about the accuracy of serum uric acid determination in predicting pre-eclampsia.

The treatment of acute infectious conjunctivitis with fusidic acid: a randomised controlled trial

BACKGROUND

Acute infectious conjunctivitis is a common disorder in primary care. Despite a lack of evidence regarding the effectiveness of topical antibiotics for the treatment of acute infectious conjunctivitis, most patients presenting in primary care with the condition receive topical antibiotics. In The Netherlands, fusidic acid is most frequently prescribed.

AIM

To assess the effectiveness of fusidic acid gel compared to placebo for acute infectious conjunctivitis.

DESIGN

Double-blind randomised placebo-controlled trial.

SETTING

Twenty-five Dutch primary care centres.

METHOD

Adults presenting with a red eye and either (muco)purulent discharge or glued eyelid(s) were allocated to either one drop of fusidic acid gel 1% or placebo, four times daily during one week. The main outcome measure was the difference in recovery rates at 7 days. Secondary outcome measures were difference in bacterial eradication rates, a survival time analysis of the duration of symptoms, and the difference in recovery rates in culture-positive and culture-negative patients.

RESULTS

One hundred and eighty-one patients were randomised and 163 patients were analysed. Forty-five of the 73 patients in the treatment and 53 of the 90 patients in the placebo group recovered (adjusted risk difference = 5.3% [95% confidence interval {CI} = -11 to 18]). There was no difference between the median duration of symptoms in the two groups. At baseline, the prevalence of a positive bacterial culture was 32% (58/181). The bacterial eradication rate was 76% in the treatment and 41% in the placebo group (risk difference = 35% [95% CI = 9.3 to 60.4]). In culture positive patients, the treatment effect tended to be strong (adjusted risk difference = 23% [95% CI = -6 to 42]).

CONCLUSION

At 7 days, cure rates in the fusidic acid gel and placebo group were similar, but the confidence interval was too wide to clearly demonstrate their equivalence. These findings do not support the current prescription practices of fusidic acid by GPs.

A randomized trial of ways to describe test accuracy: the effect on physicians' post-test probability estimates

BACKGROUND

Some people believe that likelihood ratios provide diagnostic information that is more useful than sensitivity and specificity estimates.

OBJECTIVE

To assess how physicians' estimates about probability of illness are affected by the presentation of a diagnostic test's value as an estimate of sensitivity and specificity versus a likelihood ratio or an inexact numerical graphic.

DESIGN

Random assignment of vignettes with different presentation formats of diagnostic test accuracy.

SETTING

Auditorium at a continuing medical education conference.

PARTICIPANTS

183 physicians.

INTERVENTION

After estimating probabilities of 6 common illnesses described in patient vignettes, physicians reviewed pertinent test results presented in 1 of 3 formats.

MEASUREMENTS

Physicians' probability estimates of illness before and after receiving test information, and post-test probability estimates based on the Bayes theorem.

RESULTS

Absolute percentage point differences between the physicians' estimated and the Bayes-based post-test probabilities varied from -7 to 31, from -7 to 28, and from 1 to 29 for the sensitivity and specificity, likelihood ratio, and graphical groups, respectively. Mean differences of probability estimates between the sensitivity and specificity and the likelihood ratio groups were small for all vignettes (-2 to 3 percentage points; summary mean z value across the 6 vignettes, 0.04 [95% CI, -0.14 to 0.21]).

LIMITATIONS

The small pool of participants (who were potentially selected) and the limited number of vignettes prevented a more detailed analysis of relationships between the interpreted strength of diagnostic evidence and estimations of illness probability.

CONCLUSIONS

These findings suggest that presenting diagnostic test accuracy with likelihood ratios does not affect some physicians' estimates of illness probability compared with presenting diagnostic test results as sensitivity and specificity.

Variation in requests for imaging investigations by general practitioners: a multilevel analysis

OBJECTIVES

To describe the variation in the numbers of imaging investigations requested by general practitioners (GPs) and to find likely explanations for this variation.

METHODS

Cross-sectional survey of the use of eight imaging investigations by 229 GPs collaborating in 40 local GP groups from five regions in the Netherlands during 1997. A multivariable, multilevel regression analysis was used to link these data with survey data on professional characteristics such as knowledge about and attitude towards test ordering, and with data on contextual factors such as practice type or experience with feedback on test-ordering data.

RESULTS

Data for 221 GPs (97%) were available. After adjustment for practice size and working time, the median number of imaging investigations ordered per GP was 148 (interquartile range 71 to 300), with large differences (up to twofold) between the regions (P<0.001). Overall, chest X-rays were the largest single category (median = 48 interquartile range 17 to 100). GPs working in a group practice requested, on average, 34% fewer investigations than their colleagues working in single-person practices (95% confidence interval 17 to 48%).

CONCLUSIONS

Only practice type was found to be associated with the number of imaging investigations requested, adjusted for practice size and working time factor. No further explanations were found for the inter-regional differences. Future studies on the ordering of imaging investigations by GPs should attempt to delineate contextual from regional factors.

Block design allowed for control of the Hawthorne effect in a randomized controlled trial of test ordering

BACKGROUND AND OBJECTIVE

To evaluate the value of balanced incomplete block designs in quality improvement research, and their capacity to control for the Hawthorne effect.

METHODS

General practitioners teams were randomized into three arms and received an intervention on test ordering, relating to tests for two groups of clinical problems (A tests and B tests). In the two trials within the block design, we tried to control for the Hawthorne effect by comparing the complete intervention in both arms on either the A (arm I) or B tests (arm II); the arms acted as blind controls for each other. In the classical trial, the complete intervention on B tests (arm II) was compared with a control arm without any intervention on B tests (arm III).

RESULTS

The trials with the block design yielded statistically significant changes in the numbers of A tests ordered (P=.013), but not in the numbers of B tests ordered (P=.29). In the classical design, the complete intervention reached a marginally significant change in the B tests (P=.068). The Hawthorne effect was the same for both arms of the block design. In the classical design, the effect could to some extent be attributed to the Hawthorne effect.

CONCLUSION

Our block design allowed us to control for the Hawthorne effect. Suitable use of block designs may further our knowledge of nonspecific effects in quality improvement research.

Accuracy of specific IgE in the prediction of asthma: development of a scoring formula for general practice

BACKGROUND

For the diagnosis of asthma in young children, GPs have to rely on history taking and physical examination, as spirometry is not possible. The additional diagnostic value of specific immunoglobulin E (IgE) to inhalent allergens remains unclear.

AIM

To assess the predictive accuracy of specific IgE to cat, dog, and/or house dust mites in young children for the subsequent development of asthma at the age of 6 years.

DESIGN OF STUDY

Prospective follow-up study.

SETTING

Seventy-two general practices.

METHOD

A total of 654 children, aged 1-4 years, visiting their GPs for persistent coughing (>/= 5 days), were tested for IgE antibodies by radio allergosorbent testing (RAST). Parents completed a questionnaire on potential risk indicators. Those children who showed an IgE-positive status (12.7%) and a random sample of those with an IgE-negative status (<0.5 U/ml) were followed up to the age of 6 years when the asthma status was established. The main outcome measure was asthma at the age of 6 years (combination of both symptoms and/or use of asthma medication, and impaired lung function).

RESULTS

Addition of RAST results to a prediction model based on age, wheeze, and family history of pollen allergy increased the area under the receiver operating characteristic (ROC) curve from 0.76 to 0.87. Furthermore, RAST improved patient differentiation as indicated by a change in the range of asthma probabilities from 6-75% before the IgE test, to 1-95% after the IgE-test.

CONCLUSION

Sensitisation to inhalant allergens in 1-4-year-olds, as shown by RAST, is a useful diagnostic indicator for the presence of asthma at the age of 6 years, even after a clinical history has been obtained. This model should preferably be validated in a new population before it can be applied in practice.

Coughing in pre-school children in general practice: when are RAST's for inhalation allergy indicated?

To identify patterns of clinical history associated with extreme (high or low) probabilities of allergic sensitization in coughing children so as to restrict allergy testing to those with an intermediate probability of sensitization. A total of 752 children, aged 1-4, visiting their GPs for coughing (>or=5 days), were tested for IgE-antibodies to house dust mite, cat and dog [RadioAllergoSorbent Test (RAST)]. Parents completed a questionnaire on family history of atopy, breastfeeding, smoking, pets, and floor covering. Data of 640 children could be analyzed, 83 (13%) were IgE-positive. In a logistic regression analysis, a scoring formula for the prediction of being IgE-positive was constructed using variables from the patient's history. Significant contributors for sensitization were: age (3-4 yr), infantile eczema, positive family history of mite-allergy, sibling(s) with pollen-allergy, and smoking by parents. If only one of these characteristics is present, the probability of sensitization is < 25%. In such cases watchful waiting may be preferred over allergy testing. In other cases, a negative RAST may help to exclude sensitization, whereas a positive RAST helps to establish the diagnosis. Thus, acting on clinical history alone may save approximately 80% of RAST's. Patient history-derived information contributes to distinguishing children who are at low risk for sensitization to house dust mite, cat, and dog. The scoring formula may help GPs to identify children with a low probability of being sensitized. This may form the basis for watchful waiting. In others, allergy testing may be useful to gain more diagnostic certainty.

Variation in test ordering behaviour of GPs: professional or context-related factors?

OBJECTIVE

The aim of this study was to describe GPs' test ordering behaviour, and to establish professional and context-related determinants of GPs' inclination to order tests.

METHODS

A cross-sectional analysis was carried out of 229 GPs in 40 local GP groups from five regions in The Netherlands of the combined number of 19 laboratory and eight imaging tests ordered by GPs, collected from five regional diagnostic centres. In a multivariable multilevel regression analysis, these data were linked with survey data on professional characteristics such as knowledge about and attitude towards test ordering, and with data on context-related factors such as practice type or experience with feedback on test ordering data. The main outcome measure was the percentage point differences associated with professional and context-related factors.

RESULTS

The total median number of tests per GP per year was 998 (interquartile range 663-1500), with significant differences between the regions. The response to the survey was 97%. At the professional level, 'individual involvement in developing guidelines' (yes versus no), and at the context-related level 'group practice' (versus single-handed and two-person practices) and 'more than 1 year of experience working with a problem-oriented laboratory order form' (yes versus no) were associated with 27, 18 and 41% lower numbers of tests ordered, respectively.

CONCLUSION

In addition to professional determinants, context-related factors appeared to be strongly associated with the numbers of tests ordered. Further studies on GPs' test ordering behaviour should include local and regional factors.

Predicting bacterial cause in infectious conjunctivitis: cohort study on informativeness of combinations of signs and symptoms

OBJECTIVE

To find an efficient set of diagnostic indicators that are optimally informative in the diagnosis of a bacterial origin of acute infectious conjunctivitis.

DESIGN

Cohort study involving consecutive patients. Results of index tests and reference standard were collected independently from each other.

SETTING

25 Dutch health centres.

PARTICIPANTS

184 adults presenting with a red eye and either (muco)purulent discharge or glued eyelid(s), not wearing contact lenses.

MAIN OUTCOME MEASURES

Probability of a positive bacterial culture, given different combinations of index test results; area under receiver operating characteristics curve.

RESULTS

Logistic regression analysis showed optimal diagnostic discrimination for the combination of early morning glued eye(s), itch, and a history of conjunctivitis. The first of these indicators increased the likelihood of a bacterial cause, whereas the other two decreased it. The area under the receiver operating characteristics curve for this combination of symptoms was 0.74 (95% confidence interval 0.63 to 0.80). The overall prevalence of bacterial involvement of 32% could be lowered to 4% or raised to 77%, depending on the pattern of index test results.

CONCLUSION

A bacterial origin of complaints indicative of acute infectious conjunctivitis can be made much more likely or unlikely by the answers to three simple questions posed during clinical history taking (possibly by telephone). These results may have consequences for more targeted prescription of ocular antibiotics.