General Practice

The extent of the two tier service for fundholders

BMJ 1996; 312 doi: (Published 01 June 1996) Cite this as: BMJ 1996;312:1399
  1. Robert M Kammerling, consultant in public health medicinea,
  2. Andrew Kinnear, information analysta
  1. a Avon Health, Bristol BS2 8EE
  1. Correspondence to: Dr R M Kammerling, Somerset Health Commission, Wellsprings Road, Taunton TA2 7PQ.
  • Accepted 12 April 1996


Objective: To examine possible differential changes in outpatient referrals to orthopaedic clinics, attendances, and waiting times between fundholding and non-fundholding general practitioners.

Design: Observational controlled study of referrals by general practitioners to orthopaedic outpatients between April 1991 and March 1995.

Setting: District health authority in south west England.

Subjects: 10 fundholding practices with 108 300 registered patients; 22 control practices with 159 900 registered patients.

Main outcome measures: Changes in age standardised referral and outpatient attendance ratios for the year before and the two years after achieving fundholder status; changes in outpatient waiting times.

Results: In the year before achieving fundholding status both groups were referring more patients than were being seen. Two years later, referral and attendance ratios had increased by 13% and 36% respectively for fundholders and 32% and 59% for controls, and both groups were referring fewer patients than were being seen. Attendances represented 112% of referrals for fundholders and 104% for controls. In 1991-2, a similar proportion of patients in the two groups was seen within three months of referral. The two hospitals that set up specific clinics exclusively for fundholders showed faster access for patients of fundholders by 1993-4, as did a third hospital without such clinics by 1994-5.

Conclusions: Fundholders increased their orthopaedic referrals less than did controls and achieved a better balance between outpatient appointments and referrals. Their patients were likely to be seen more quickly, particularly if the hospital provided special clinics exclusively for fundholders. Lack of case mix information makes it impossible to judge whether these differences benefit or disadvantage patients.

Key messages

  • Patients of fundholders had no better access than patients of non-fundholders to orthopaedic services before the practices became fundholding

  • Fundholders controlled their referrals better than non-fundholders and achieved a better balance between referrals and attendances

  • Fundholders' patients were more likely to be seen quickly, especially if the receiving hospital laid on specific clinics exclusively for fundholders

  • Information on case mix is needed to identify whether the lower referral rate among fundholders benefits or disadvantages patients


Since general practitioner fundholding was introduced, its ability to produce a two tier service has been described,1 occasionally denied,2 and eventually accepted as inevitable.3

Changes in referral patterns for general practitioners who became fundholders in the first year have been reported.4 5 Although there is increasing emphasis on reducing the waiting times for outpatient appointments,6 there is little hard information about the effects of fundholding on this. A recent review of the first three years of fundholding did not present any data,3 and where data have been presented they have focused on a single practice.7

We used routine data to look at the effects of fundholding on general practitioner referrals, outpatient attendances, and speed of access to outpatient services in Bristol and District Health Authority. This served a total population of 840 000, of whom 15.4% were registered with the first three waves of fundholders. We concentrated on orthopaedics, as this had been identified as the top priority for improvements in access in a survey of general practitioners that coincided with the start of the fundholding initiative.8


Routine data were available to the health authority on all its local residents (whether registered with a fundholder or not) who used outpatient services from 1 April 1991 onwards—when the NHS reforms were introduced. These data included, but were not limited to, the contract minimum data set, and contained the age and sex of people seen at outpatient clinics, whether at a first or follow up appointment, the date of referral, the hospital of referral, the date of attendance, and whether a general practitioner referral and if so the practice, with a specific code for fundholders. Similar information, other than details of the actual attendance, was available for people still awaiting an outpatient appointment.

To examine the changes before and after becoming fundholders, we examined second and third wave fundholders only (the fundholding group), as we had no information on first wave fundholders before they became fundholding. We also identified a control group of practices who were not, and still have not become or applied to become, fundholders. Each practice was matched to two control practices that used the same hospital of main referral and served areas of similar socioeconomic status. In two cases, a third control practice had to be added to ensure a broadly similar age distribution between the fundholding practice and its control population.

We identified first referrals by general practitioners, first outpatient attendances, and waiting times for both groups for each financial year from 1 April 1991 to 30 March 1995. We calculated expected numbers of referrals and attendances for each year by age group, using the whole district rates for 1992-3. This enabled us to calculate the standardised referral and attendance ratios by combining data from different years so as to arrive at a single value for the year before fundholding and the two subsequent years, even though the practices became fundholders in different years. Analogous calculations were carried out on the control group, treating each practice in the same way as its matched fundholding practice.

Patients' waiting times were compared on a calendar year basis, from 1991-2, before any practices in the fundholding group becoming fundholders, to 1994-5, when all practices in the fundholding group had been fundholders for at least two years. We used the proportion of referred patients seen within three months of referral as an indicator of speed of access.


The fundholding group consisted of 10 practices with 108 300 patients. The control group had 22 practices with 159 900 patients.

In the year before becoming fundholders, the practices in the fundholding group were referring patients at a lower rate than the controls, although the difference was not significant, and first attendance rates were similar in the two groups (table 1). By the second year after achieving fundholding status, fundholders had increased their referral rates by 13% (ratio of standardised ratios 1.13; 95% confidence interval 1.03 to 1.23) and the controls by 32% (1.32; 1.23 to 1.43). Outpatient attendance had increased by 36% (1.36; 1.28 to 1.49) for the fundholders and by 59% (1.59; 1.49 to 1.72) for the controls. In the year before fundholding, the number of attendances was 93% of the number of referrals for fundholders and 86% for controls; two years later, attendances had increased to 112% and 104% respectively.

Table 1

Changes in outpatient referrals and attendances

View this table:

Patients' waiting times were compared for each of the four hospitals providing orthopaedic services. In 1991-2, the proportion of patients seen within three months was similar for the two groups of general practitioners but differed among hospitals (table 2).

Table 2

Changes in percentage of patients seen within three months of referral over four years, by hospital

View this table:

Two hospitals (A and D) set up specific clinics exclusively for fundholders. By 1993-4, fundholders' patients were more likely to be seen quickly at these two hospitals. In 1994-5, fundholders' patients were also more likely to be seen quickly at one of the other hospitals.


To assess the possible differential changes in orthopaedic outpatient referrals, attendances, and waiting times between fundholding and non-fundholding general practitioners we used routine information, which was also used for contract monitoring and to illuminate specific problems. These data had therefore been subject to substantial scrutiny and we believe them to be generally accurate.

Specific initiatives to reduce outpatient waits took place during the period studied. These concentrated on ensuring appointments for patients who had been waiting the longest, with the inevitable consequence that more people who had waited longer were seen. We reanalysed the data excluding long waiters to counteract this situation, but this made no significant difference to the results.

There was a general increase in access to orthopaedic outpatients during the period studied. In the year before achieving fundholding status, practices referred more patients than outpatient slots were available, but at the end both fundholders and non-fundholders had more outpatient attendances than referrals. This difference was greater for fundholders, and this was reflected in their patients' shorter waiting times. Fundholders also referred fewer patients than non-fundholders, and this difference increased over the study period. This is different from the experience in the Oxford region,5 where first wave fundholders had a higher referral rate in the year before becoming fundholders.

As the diagnostic information on outpatients is not contained within the routine data sets, we do not know about the case mix of referrals and we therefore cannot comment on whether the differences in referrals were related to changes in the threshold of referral between the two groups. One of the hospitals offered no separate service for fundholders' patients, but they were still more likely to be seen quickly there in 1994-5. It is possible that fundholders referred only the sicker patients, who were thus allocated a higher priority and therefore were seen more quickly than other patients.

Although these results do show a significant effect of fundholding on general practitioners' referral rates and speed of access to services, they do raise the question of whether these changes actually benefit patients. It is possible that the downward incentive of fundholding on referrals results in the referral of patients only when their problems become more severe. The appropriateness of such a response is still open to doubt.

We thank Mike Shepherd for his hard work in identifying the socioeconomic status of areas served by local general practices, and the anonymous assessor who insisted on the use of a control group.


  • Funding None.

  • Conflict of interest None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
View Abstract