Determinants of GP billing in Australia: content and time

In 1989, the Medicare Benefits Schedule (MBS) for general practice A1 attendances in Australia changed, and content-based consultation descriptors were introduced with four levels of consultation (Box 1).1 Level A and level B consultations are distinguished only by content, regardless of how much time is spent with the patient. However, level C and level D consultation descriptors have both minimum time and content requirements.2

If content requirements are met but the consultation is shorter than the specified time (level C ≥ 20 minutes; level D ≥ 40 minutes), or if time requirements are met but not content, then the item can only be reimbursed by Medicare at the lower rate.3 General practitioners may “down-code” (eg, claim level B for a consultation of 30 minutes) on the basis of content, but cannot “up-code” (eg, claim level C for an 18-minute consultation) on the basis of content.

There has been little research into GPs’ application of this complex system. One study reported the application of content descriptors by having GPs observe and rate videotapes of GP consultations.4 Others looked at the qualitative content of consultations, but did not consider the application of the MBS descriptors.5,6 There have been no studies of GPs’ consideration of time and content when deciding which Medicare item number to allocate to a consultation.

The current Medicare descriptors make it clear that GPs must consider both time and content in choosing an item number. However, in 2003, the Attendance Item Restructure Working Group, which consisted of members from GP organisations and the Department of Health and Ageing (DoHA), reviewed the A1 Medicare attendance items. Although the GP organisations preferred to retain a mixture of content and time in descriptors, the DoHA preferred a seven-tier item structure based on time alone.7

If GPs are currently applying both content and time criteria in item number selection, modelling future costs of a seven-tier time-based item structure on the basis of current MBS claims data will provide an inaccurate cost estimate.

We examined the extent to which content relates to consultation length and Medicare itemisation, and examined factors influencing a GP’s decision to bill a level B consultation when time requirements for level C are met.

We conducted a secondary analysis of a subset of data from the Bettering the Evaluation and Care of Health (BEACH) study, a continuous national cross-sectional survey of general practice in Australia. About 1000 GPs participate each year, each providing details (on structured paper encounter forms) about 100 consecutive patient encounters. Data include payment source and, where applicable, Medicare item number.8

From 1 April 2000 to 31 March 2003, each GP was asked to record consultation start and finish times for 40 of their 100 encounters, samples being randomly placed in each recording pack (first 40, middle 40 or last 40). For about 200 GPs, these data elements were included on 70 forms. Regular data-cleaning procedures checked outliers in calculated consultation length for data entry error and for sequential time-fit of encounters. Impossible data were eliminated. Problems managed at encounter were classified according to the International Classification of Primary Care (ICPC-2).9 Chronic problems were selected using the ICPC-2 rubric groupings of O’Halloran et al.10

We examined four consultation groups:

• encounters ≤ 20 minutes designated by the GP as Medicare level B.

• encounters > 20 minutes designated as level B or level C, and two subsets:

• encounters > 20 minutes designated Medicare level B;

• encounters > 20 minutes designated Medicare level C.

We compared place of consultation, characteristics of patients seen and consultation complexity (measured by number of reasons for encounter, number of problems, number of chronic problems, rates of medication, clinical treatments, pathology and imaging tests ordered, and referrals made). The distribution of problems managed across ICPC-2 chapters was also compared.9

There were 80 476 level B consultations ≤ 20 minutes, 5725 level B consultations > 20 minutes, and 9168 level C consultations > 20 minutes. These level C consultations were significantly but marginally longer (29.2 minutes) than the longer level B consultations (28.3 minutes) (Box 2).

We could not validate recorded start and finish times. A previous study showed a large proportion of consultations recorded as exactly 10, 15 or 20 minutes, suggestive of rounding, which casts some doubt about the study’s accuracy for assessing exact length of consultation.12

The Medicare requirements for a level C claim include both time and content. If time were the prime factor in GP item selection, we would expect a considerable difference in duration of the level C and longer level B consultations. However, the difference was somewhat marginal, with a difference of less than 1 minute for the means and 3 minutes for the medians. Clearly, factors other than time influence GP item choice.

Patient age and Veterans’ Affairs card status are clearly related to the length of the consultation, but do not relate to the GP’s choice of item number. In contrast, patient health care card status relates to choice of level B or C in longer consultations, but not to consultation length per se. Health care card holders have been shown to have a greater morbidity burden than other general practice patients,13 and morbidity and age are closely linked, so these factors may be reflecting morbidity patterns rather than being independent predictors of either length of consultation or item choice. However, GPs may be less inclined to charge a patient without a health care card for a level C consultation (if they do not bulk bill them), because of the cost to the patient.

The higher proportion of female patients at longer consultations and the even higher proportion at level C consultations may reflect a mixed effect of morbidity and GP sex: women more often see female GPs,14 whose consultation times are, on average, longer than those of male GPs,12 and female patients have higher rates of psychosocial and genital problems.14

The greater proportion of patients new to the practice at longer consultations, and especially at the level C consultations, suggests that new-patient status may be independent of morbidity in predicting length and item number, as a more complex history is required. Specialist physicians have always charged more for the first consultation. There is no allowance for this in the MBS for general practice, so GPs may be more likely to charge a level C consultation for new-patient assessment.

GPs appear to be using both time and content appropriately in choosing item numbers. The differences between longer consultations billed as level B and those billed as level C indicate that considerations other than time influence billing decisions. Moreover, this does not appear to be simply down-coding (where a GP who has fulfilled the descriptor bills a lower-value item number). There appear to be valid reasons for apparent down-coding (ie, GPs appear to have a strong sense of the relative value of consultation content).

More than a third of the BEACH longer consultations were designated level B. The 2003 Medicare claims data show that practitioners claiming A2 items of service (which do not require consideration of content) billed 15% of all claimed attendances as long consultations, whereas those billing A1 items claimed 11.9% at level C.15 If the longer level B consultations in BEACH are added to the level C consultations in BEACH, longer consultations represent 14.7% of all A1 items of service, which aligns with the pattern of A2 claims. This suggests that if future Medicare items are purely time-based, “bracket creep” will result and the pattern of claims will change, so that modelling future costs on the basis of current Medicare claims will not generate reliable cost estimates.

We plan to further investigate the relative importance of these factors and of GP characteristics in determining choice of level B or C for longer consultations. Meanwhile, this study may provide an opportunity for more educated discussion on future Medicare fee structures for GP attendances.