Skin checks for potential skin cancers in general practice in Victoria, Australia: the upfront and downstream patterns and costs

Source data

This is a retrospective cohort study using the subcomponent of the MedicineInsight longitudinal general practice database covering the state of Victoria.¹⁰ The database contains de-identified records for each patient consultation/encounter (“encounters”) from the clinical information systems at participating practices (“sites”), including patient demographics, encounter details, and clinical/management data entered by practice staff. Each record can include multiple encounter reasons, entered via a drop-down menu and/or free-text field. The study cohort comprised people in Victoria with a record indicating a skin check as an encounter reason (Supplementary Table 1, available from: figshare.com/articles/journal_contribution/_/25727706).

For completeness of cost information, analyses were restricted to practices with integrated billing data, including MBS claims – this accounted for > 70% of all patients. Multiple records for an individual at the same practice could be identified in the MedicineInsight dataset using a unique patient number, and prescribed medications were linked using a unique patient encounter number. BioGrid Australia, a connectivity platform for medical data¹², probabilistically linked records across sites when sufficient identifiers (for example, name, address, date of birth) were present. The proportion of encounters that had an encounter reason recorded varied by site (5–95), and we excluded five sites that had < 30 of encounters with an encounter reason recorded (See Supplementary figure 1, available from: figshare.com/articles/journal_contribution/_/25727706).

Healthcare

We analysed skin check encounters and related healthcare that occurred in the 3 months after the encounter, as many treatments would be scheduled for another day. The study period was from January 2010 to December 2017. To allow 3 months of follow-up, we considered skin checks up to September 2017. Demographic information included age, gender, socioeconomic status (using the Index of Relative Socio-economic Advantage and Disadvantage¹³) and remoteness of place of residence.¹⁴

We identified skin cancer–related care using MBS item numbers specific to melanoma, keratinocyte carcinomas and benign or other skin lesions (Supplementary Table 1, available from: figshare.com/articles/journal_contribution/_/25727706). The MBS items for skin cancer treatment required histological confirmation of malignancy, by type, to qualify for reimbursement. Other treatments were identified using MBS items and/or encounter reasons. Skin-related medicines were identified from medicine-specific tables in MedicineInsight by searching the medicine name or active ingredient (Supplementary Table 1, available from: figshare.com/articles/journal_contribution/_/25727706). Our focus was on skin checks related to potential skin cancer and subsequent skin-related healthcare, so any records missing the encounter reason and not identified in the skin cancer–related care items were not included in the final analysis.

Costs in general practice

Health system costs (government-as-payer perspective) were assigned to each MedicineInsight encounter according to the rebate payable for the MBS items listed on that date. Only GP consultation and skin-related MBS items were included in costing. The costs of skin-related medicine prescriptions were assigned using subsidies listed in the PBS. All costs were converted to 2020 Australian dollars using the Australian health index from the Australian Consumer Price Index.¹⁵

Statistical analysis

Each skin check encounter was categorised by consultation length (from the corresponding MBS item) and demographic information. We analysed the proportion of patients who had another skin check or other related healthcare up to 3 months after the initial skin check, and the associated costs. We investigated the medicines prescribed, MBS-indicated diagnoses from skin excisions (keratinocyte carcinoma, melanoma, benign/other), and frequency of skin checks. We presented time and cost results as a mean, median, standard deviation (SD) and interquartile range (IQR), and used a sensitivity analysis to analyse treatment records up to 1 month and 6 months after the intial skin check.

Ethics and data storage

Ethical approval was granted by Melbourne Health Human Research Ethics Committee (202002/1) and MedicineInsight Governance Committee (2020-020). Study data were stored within a secure research environment at the University of Melbourne, with access restricted to authorised users. Analyses were carried out using Stata v16.

Skin checks

For the period between January 2010 and September 2017, we identified 59 046 skin check encounters for 40 014 patients across 73 general practice sites. Skin checks represented 0.7% of all encounters with a “reason for encounter” recorded. Of the 73 sites identified, 51 (70%) were in major cities and 18 (25%) were in the highest socioeconomic quintile areas. The mean skin checks per practice was 809 (median 352, IQR 141–774), with 16 skin checks per practice per month (median 7, IQR 3–19). The median age at a skin check was 52 years (IQR 36–67). Most people (65%) were from major cities and 29% were from the highest socioeconomic quintile areas (Table 1). There was an increasing number of skin checks each year due to the increasing coverage of the MedicineInsight dataset during the study period, with additional practices being included in the dataset each year (See Supplementary Table 2, available from: figshare.com/articles/journal_contribution/_/25727706).

Among the people having a skin check, there was a mean of 1.5 skin checks per person (SD 1.1, IQR 1–2) and 26% had multiple skin checks within the study period. The mean follow-up per person was 4.1 years (SD 2.4, median 4.3, IQR 2.1–6.2) between the first and last recorded encounters of any kind. Among the people having multiple skin checks, 28% of subsequent skin checks were within 3 months of the original skin check, 32% were after > 3–12 months, 25% were after > 1–2 years and 15% were after > 2 years. Among the subset of 10 875 patients with a BioGrid (cross-site) identifier, 0.6% had a skin check at multiple sites.

Grouping the skin checks by their encounter reason, 38% were recorded as being for a skin lesion or damage, 36% were recorded as a skin check or exam, 18% were recorded as being for a mole or naevus and 8% were recorded as skin cancer screening. Ninety per cent of skin check encounters were billed as GP consultations in the MBS: 65% were standard GP consultations of < 20 minutes (decreasing from 71% in 2010 to 55% in 2017), 23% were long GP consultations of 20–40 minutes (increasing from 18% in 2010 to 32% in 2017) and 2% had time groupings from other MBS GP consultation items. Seven per cent of skin check encounters had no MBS claim entered on the same date, and the remaining 3% had a non-consultation item (for example, management planning) or an invalid MBS item number.

Downstream management

A skin biopsy and/or skin excision was recorded ≤ 3 months after 18% of skin checks (Table 2), with 17% of excisions and 21% of biopsies performed on the same day as the skin check. There was a record of a skin-related topical medicine after 2% of skin checks, and 5 had other skin-related treatment (90% cryotherapy). Overall, at least one of these treatment items was recorded in the 3 months after 20% of skin checks; it was within 1 month for 18% of skin checks and within 6 months for 21% of skin checks, the sensitivity analysis found. The proportion of skin checks with subsequent care appeared to be higher with older age and for people from outer regional and more disadvantaged areas (Figure 1). For 9% of skin checks, the patient had another skin check within 3 months and for 4% of checks, patients had another skin check after 3–6 months, suggesting that lesions were being monitored for change over this period. Of the people who had another skin check within 3 months, 37% had a biopsy and/or excision in the subsequent 3 months.

Figure 1.

Proportion of skin checks with subsequent skin-related healthcare within 3 months and mean general practice-related health system costs over that time for those having subsequent care

MBS items indicating the histopathological diagnosis of an excised or biopsied lesion were claimed (within 3 months) after 11% of skin checks – 2% of these indicated melanomas, 29% were keratinocyte carcinomas and 68% were benign or other skin lesions. The proportion of skin checks with a subsequent excision ranged from 8% for people aged < 30 years to 17% for people aged ≥ 80 – the main age-based difference was for keratinocyte carcinomas (< 1% and 11%, respectively). The proportion of skin checks that had a subsequent biopsy was 4% for people aged < 30 years, 13% for people aged ≥ 80 years and 19% for people from outer regional areas (Supplementary Table 3, available from: figshare.com/articles/journal_contribution/_/25727706).

Costs of skin checks and downstream management

Of the 54 875 skin checks with valid MBS billing data on the date of the encounter, 94% were billed as a GP consultation only, 5% had MBS billing codes for a GP consultation and other skin-related healthcare, and 0.5% only had an MBS billing code for skin-related healthcare (for example, skin biopsy). The mean cost of a skin check encounter was A$64 (SD 38, median 50, IQR 44–80). For skin checks without subsequent skin-related healthcare, the mean cost was A$58 (SD 24, median 49, IQR 43–77), with little cost variation across patient characteristics (Supplementary Table 4, available from: figshare.com/articles/journal_contribution/_/25727706).

For skin checks with subsequent skin-related healthcare (including biopsy or excision on the same day), the mean total costs (including skin-related healthcare) in the subsequent 3 months was A$240 (Table 3). Costs varied based on the type of healthcare, ranging from A$162 for people who had a biopsy without a separate excision to A$595 for those having surgical excision of a melanoma (Table 3). Costs increased with increasing age. While the proportion of people with subsequent skin-related healthcare varied markedly according to other characteristics (gender, remoteness and socioeconomic status), there was less variation in costs across these characteristics. (Figure 1; Supplementary Table 5, available from: figshare.com/articles/journal_contribution/_/25727706).

Limitations

This study has several limitations. The reason for each encounter in the MedicineInsight dataset is recorded in a free-text field, so we may not have identified all skin check encounters if they were not clearly recorded. We were also unable to determine whether the encounter was a routine skin check, a patient-presented lesion or mole of concern, or an incidental finding during a general consultation. In Australia, approximately one-third of people report some type of skin check in the past 12 months, with 22 reporting a whole-body check (37 for people aged 45–69 years).²⁰ Moreover, MedicineInsight is primarily a quality improvement program, so it is possible that the included GP practices are not representative of all GP practices. However, the sample is considered to be broadly representative of people who visit GPs, and the relevant data are generally well recorded.^22,23

In addition, we could not determine if the included sample had additional skin checks in other practices that were not part of this study, potentially underestimating their healthcare use and costs. However, we checked a subset of our sample who could be linked across multiple practices and found there was limited use of multiple practices. Also, we relied on MBS item numbers to indicate skin cancer diagnoses, rather than “gold standard” cancer registry data.

Further research could address patient out-of-pocket costs associated with these encounters and downstream care, especially with falling rates of general practice encounters in which patients are not charged additional out-of-pocket fees above the Medicare rebate (i.e. bulk-billing).²⁴ The variation in frequency of subsequent healthcare after a skin check, especially for people from outer regional and more disadvantaged areas, suggests that access to skin checks may not necessarily reflect need. However, additional patient information is needed to make stronger inferences about whether this represents the inverse care law – that people most in need of healthcare are least likely to receive it.