Summary

Objective: Literature shows trainers have lower referral rates than trainees without providing detailed explanation. This paper aims to study the agreement between trainers and trainees on the appropriateness of referrals made to the emergency department.

Design: A retrospective study (qualitative and quantitative) of referral letters. A public outpatient clinic, that is also a training centre for community-based family medicine vocational training in Hong Kong.

Subjects: Three groups, namely: (i) five trainers; (ii) six trainees in first six months of training (C1); (iii) six trainees after two years of community-based training (C2).

Main outcome measures: Comparison of ratings on agreement or disagreement with the appropriateness of referrals.

Results: Within the groups, the ratings are highly congruent. However, the trainers disagree with 12.3% of the referrals, but the C1 trainees disagree with only 3.9% while the C2 disagree with only 7.3%. The disagreement has no relationship with the specialty or organic system referred, the referring doctor, or the purpose of referral.

Conclusion: Trainers are more likely to disagree with the referrals made than the trainees. Trainees should discuss referred patients with their trainers as part of their learning process.

Keywords: Referral, family practice, emergency hospital service, vocational education

摘要

目的：文獻顯示家庭醫學導師比其學員較少轉介病人，但無詳細解釋原因。本文旨在研究導師與學員對病人被轉介到急症室的恰當性，意見是否一致。

設計：在一個設有家庭醫學培訓的公立門診部門，對病人轉介信的內容和數量進行回顧性研究。

對象：執業醫生，分開三組，包括五名家庭醫學導師、六名正接受首六個月訓練（C1）的學員、及六名已完成兩年社區基礎培訓學員（C2）。

測量內容：將各組別對轉介病人到急症室恰當性的評級作比較。

結果：在同一組別，評分等級極為一致。然而，導師不同意12.3%的轉介，C2組別不同意7.3%的轉介，C1組別則不同意3.9%的轉介。意見的分歧和被轉介的專科組別、作出轉介的醫生和轉介的目的並無關係。

結論：導師比學員較多對病人被轉介到急症室持不同意見。學員應該在學習過程中跟導師討論病人的轉介。

主要詞彙：轉介，家庭醫學，急症醫療服務，在職教育

---

Introduction

Family physicians refer patients to the accident and emergency department (AED) for various reasons: patients' conditions, limited diagnostic and/or treatment facilities available in the primary care setting, unresolved requests from patients, etc. It has been shown that the referral rates and patterns differ among different practices in and outside of Hong Kong.^1,2 A large survey of referral patterns in local family practice¹ involving 15146 consultations (317 referrals) revealed that the referral rates varied widely from 0.4% to 7.8% and this was not related to the years of experience as a family practitioner. Referral to AED comprised 3.8% of all the referrals.

From a search in PubMed, only a few reports in the English language could be found on the factors that influence the referral rates in family practice. One Finnish study on 2921 referrals by 851 health centres found that family physicians' professional experience and specialist training were negatively related to referral rates.³ But another survey in the UK showed that the referring family (general) practitioners with particular expertise in ophthalmology and otorhinolaryngology had higher referral rates to these specialties. The eventual conclusion was that "a high referral rate does not necessarily imply a high level of inappropriate referral".⁴

There were only two studies found which compared the behaviour of trainers and trainees. One showed that trainers had a lower referral rate when compared with trainees for out-of-hour service provision.⁵ The other revealed that trainees tended to follow their trainer' s referral patterns.⁶ From contacts with trainees and trainers, the authors noted that trainees and trainers very seldom discuss patients who were referred out. This appears to be an area of vocational training which might not have received the deserved attention.

We postulated that trainers did not agree with the trainees on some of the referrals made to AED. The aim of this study is to test out this hypothesis in a public general outpatient clinic that is also a training centre for community-based family medicine vocational training in Hong Kong. AED was chosen because patients could attend it without referral and thus doctors' referrals were unlikely to be influenced by patients' unresolved requests.

Method

All the referral letters to AED during the period October 2002 to July 2003 inclusive were collected retrospectively. These letters were computer printouts. They were photocopied and the names of the referring doctors and patients were covered to maintain confidentiality. Three groups of doctors were then asked (five trainers, six trainees in their first six months of community training (C1), and six trainees who have finished their two year community training (C2)) to rate the referrals and see if they agree with the appropriateness of the referrals or not, using a four-point scale: 1 - agree, 2 - doubtfully but tend to agree, 3 - doubtfully but tend to disagree, and 4 - disagree. "Appropriateness" was taken as that "the patient under consideration should be referred". They were blinded to the hypothesis of the study and the ratings of others.

The raters were encouraged to make comments on the referrals. These comments were then grouped under different themes and presented qualitatively.

Cronbach alpha was used to test the congruence of the ratings within each group of doctors. The ratings for each referral within a single group were averaged to get the mean rating for that referral for the group. The means between trainers and C1 trainees, between trainers and C2 trainees, and between trainees, were then compared using the paired sample t-test. Analysis of variances (ANOVA) was used to assess the relationship between ratings and other independent variables. The p value of less than 0.05 was taken as statistically significant.

Results

There were 137 referrals made to AED (136 patients: 58 males, 77 females and one missing value) within the study period by 22 doctors. All the referrals were initiated by the attending doctors, none on patient' s request. Only one referral was by a trainer while the other doctors were in different stages of their vocational training, including one in higher training. So, trainees made virtually all the referrals. The mean age of the patients was 57.523.5 years, ranging from one month to 93 years. The mean rating with standard deviation for the trainers was 1.720.71 (1 = agree, 2 = doubtfully agree). The corresponding means for C1 and C2 trainees were 1.580.55 and 1.590.64 (Table 1). The Cronbach alpha values were high for the three individual groups.

Paired t-test shows that there is no statistical significance between the trainees but a significant difference between the trainers and the trainees (between C1 and C2 trainees t=0.459 p=0.647, between trainer and C1 trainees t=3.501 p=0.001, between trainer and C2 trainees t=3.451 p=0.001). The trainers disagreed more on the referrals (Table 2). While the trainers definitely agreed with 59.8% of the referrals, they disagreed with 12.3%. The trainees disagreed with 3.9-7.3% of the referrals but agreed with 62.4-65.2% of them.

The ratings of the three groups have no significant relationship with (a) the referring doctor (ANOVA F=0.761 p=0.52) (b) the specialty of the referred pathology (ANOVA F=0.674 p=0.57), or (c) the system pathology for which referral is made (ANOVA F=0.824 p=0.48).

The comments by the raters who disagree with the referrals are grouped under four themes and some of the comments are quoted as follows (the diagnoses of referral are in brackets).

(a)		Inadequate documentation
		-		Poor description of the location of the foreign body (foreign body in eye)
		-		Not mention of diagnosis and prescription (chest pain / palpitation)
		-		No mention of breathing difficulties (motor neurone disease, deterioration
		-		? Suspected unstable angina or atypical chest pain, not mentioned in referral letter (chest pain)
		-		Inadequate information about the severity from note (localized skin infection)
		-		Symptoms and signs not well documented (high blood pressure, ? temporal arteritis)
		-		History not detailed enough (acute blurred vision)
(b)		Inadequate management
		-		Should be able to detect papilloedema [instead of referral] (? malignant hypertension)
		-		Suggestive of compliance problem, should not be managed by emergency department, no symptoms (uncontrolled hypertension)
		-		Need to explore medication (chest pain)
		-		Increase anti-hypertensive drug, mobilize other social support (hypertension with social problem)
(c)		Inadequate indication for referral
		-		Fever for one day, 39.2, general condition good (fever for investigation) - Cough for 3 weeks, general condition good, no chest sign (asthma)
		-		Only increase menses for 3 days (menorrhagia)
		-		No indication (hyperglycaemia)
		-		No emergency, fever seemed to be subsiding (fever for investigation)
		-		Condition stable (chest pain)
		-		General condition good, can observe (fever for investigation)(d) The patient could be managed otherwise without referral
		-		May be treated in general outpatient department with early follow-up (eye infection)
		-		Treatment in outpatient department (infected sebaceous cyst)
		-		Treat and review next day (eye abrasion)
		-		Need adjustment of medication (angina pectoris)
		-		For observation if parent agrees (foreign body in throat)
		-		Refer specialist outpatient [instead of to emergency department] (anaemia)
		-		Arrange community nurse service or home-helper (hypertension with social problem)
		-		Why not urgently refer eye? (acute blurred vision)

Discussion

In this study, the trainers disagreed more than the trainees on referrals to the hospital emergency department initiated by trainees. The trainers agreed with the trainees on 60% and disagreed on 12% of the referrals; the corresponding proportions by the trainee-peers were 62-65% and 4-7% respectively. The trainers were doubtful on the appropriateness of 28% of referrals. Within their individual groups, the trainers and trainees were highly congruent in their ratings of the referrals, and their ratings had no significant relationship with the referring doctor, specialty to be referred to, or organ system of pathology.

The previous surveys^1-3 noted the different referral rates among practices. The present study noted the different opinions on the appropriateness of referrals within a single practice as rated by trainers and trainees. Because of the small number of trainers, it is not possible to assess the influence of years of experience. Judging from the high Cronbach alpha within the group in spite of a large range of years after obtaining the secondary qualification in family medicine (from 2 years to 18 years), the influence of years of experience is probably small. With this assumption, training in family medicine to the specialist status may affect the appropriateness of referrals.

Training appears to improve the quality of referrals. Though the C2 trainees gave overall ratings not significantly different from those of C1 trainees, the disagreement with the referrals was between the C1 trainees and the trainers (Table 2). Moreover, C2 trainees had essentially the same doubtful ratings (doubtfully agree and doubtfully disagree) as the trainers, 27.5% and 27.8% respectively while the C1 trainees were doubtful on 33.7% of the referrals. Whether training rather than years of experience is more important in referral pattern needs further research.

It should be noted that this study does not assess the referral rates by the trainers. That the trainers do not agree with the appropriateness (whether the patient should be referred) does not necessarily mean that they have a lower referral rate. At best, they can be expected to refer less by selecting the appropriate patients. If the assumption that trainers have better selection is correct, this offers an explanation for the lesser referral rate by trainers observed in other studies. From the comments made by the raters (trainers and trainees), some referrals are not and cannot be supported by documented evidence. The validity of the ratings is hence limited by the quality of the documentation in the referral letters. A fairer assessment can be made by viewing videotapes of the actual consultations but is extremely difficult to carry out.

This study is further limited by being a survey in a single outpatient clinic and by the small sample size of trainers and trainees. We hope that similar studies would be replicated, with the implication that training improves referrals.

As there is a difference between trainers and trainees in referral decisions (dis-agreement on 12% and doubts on 28% by the trainers), there should be arrangement for discussion regarding those patients who had been referred to secondary care: diagnosis, management, indication, reasons for and expectations from the referrals. This should be part of the learning process of vocational training.

Key messages

1. In a community-based training centre for family medicine, trainers disagreed definitely with 12% of the trainees' referrals to the hospital emergency department.
2. The trainees at different stages of vocational training disagreed definitely up to 7% of their peers' referrals.
3. There was no association between the disagreement and the referring doctor, the specialty to be referred to, or the organ pathology of the referrals.
4. The main reasons for disagreement were no urgency in the referral and the feasibility of management in the community.
5. Trainees and trainers should arrange opportunities to discuss the trainees' referrals.

---

---

References