Inclusion and exclusion
The 2009 database of CZ, a large Dutch healthcare insurance company (n=3.419.604 insured persons, approximately 21% of the Dutch population) was
retrospectively analysed. Insured patients that had received an invoice related to the diagnosis "PAD Fontaine 2" corresponding to IC complaints as diagnosed by a vascular surgeon, were eligible for inclusion. Only patients that had been insured for at least two consecutive years at CZ were eligible, excluding crossover patients from other insurance companies (possibly harbouring an unknown 2-year history of PAD). In order to restrict the study to newly diagnosed IC patients, all patients that underwent vascular interventions in 2007 or 2008 were also excluded (figure 1). Data on comorbidity (diabetes mellitus, COPD, hypercholesterolemia and heart failure) were collected on the basis of prescribed medication.
Definition of IC subgroups
Patients meeting inclusion criteria were subdivided into three groups based on their primary treatment as initiated by a vascular surgeon within three months after diagnosis (t0< t 3months) (figure 1).
- SET group: community-based SET started in the period between 12 months prior to diagnosis up to 3 months after the diagnosis of IC by a vascular surgeon (graph 1). SET prior to diagnosis was possible in case of referral by a patient's general practitioner (GP) who already initiated SET.
- INT group: Any form of revascularization received within 3 months after the diagnosis of IC by a vascular surgeon.
- REST group: Neither SET nor INT group inclusion criteria were met (graph 1). It was assumed that these patients received a "go home and walk" advice or had no IC at all (this contradictory 'no IC in an IC population' phenomenon is explained later on in the sensitivity-analysis paragraph).
Secondary treatment was defined as a second intervention (primary at other location and/or re-interventions or SET) that was performed following primary treatment. This population included patients with either failure of primary treatment (re-intervention), treatment of the contralateral leg as well as treatment delay (treatment > t3months). Differentiation between these groups was not possible due to the nature of the database. It was assumed that all patients received best medical treatment (BMT) prescribed at the discretion of the physician.
Costs of PAD treatment
A Fontaine classification by a vascular surgeon is always required for billing purposes in the Dutch healthcare system. All claudication related invoices (by physician and physiotherapists) for primary as well as secondary treatment within two years of follow-up after t0 were screened. The majority of interventions in 2009 were performed by radiologists, based on a vascular surgeons indication. Total costs generated in the SET, INT and REST groups were calculated by adding all intervention costs for the group the patient was initially allocated to. Discrimination between a secondary ipsilateral intervention and a primary contralateral intervention was not possible due to the nature of the database, aimed at costs registration. Mean total costs (MTC) per patient per
group were calculated. Expenditures for medication use were not incorporated into the calculations.
Sensitivity-analysis of the REST group
Patients in the REST group were supposed to have received a walking advice. When a patient is referred to a vascular surgeon for IC by a GP, a "PAD Fontaine 2" (synonymous to IC) invoice is registered. However, if IC is subsequently ruled out, the invoice is often not corrected into the proper invoice "ruling out IC". In a prospective analysis of 100 consecutive patients we found that 30% of the GP referrals with presumed IC in fact received an alternative, non-IC diagnosis (21). This phenomenon is not financially driven since reimbursement of a "rule out" invoice appears €1.43 higher. This incorrect registration may have contaminated the composition of the REST group and it was decided to exclude the REST group from the SCM cost-analysis. However, a cost-analysis of the REST group was performed in an additional sensitivity analysis (SA), assuming 30% non-IC patients.
Cost-analysis of a nationwide adherence to a SCM
The costs of a SCM were estimated for patients in the SET and INT group. Successful treatment guided by a SCM depends on two critical success factors:
- the surgeons' compliance (willingness) to refer each IC patient to SET. Despite the fact that, according to contemporary international guidelines, all patients with IC initially should be referred for SET (2-4), this ratio was hypothetically set to 80% (best), 50% (moderate) and 30% (worst) based on reimbursement issues, appreciation of SET by vascular surgeons and preference (although evidence to do so is lacking) for a revascularization in certain cases.
- the patient's motivation to participate in a SET program, which largely depends on reimbursement issues (level of compensation by the insurance company) as well as a thorough understanding of the benefits of a SET program in comparison to an invasive intervention. The latter is also strongly associated with the surgeon's knowledge regarding SET and the willingness to refer. To allow for a variation in patients' willingness, this ratio was arbitrarily set at 80% in case of full reimbursement for all IC patients and optimal provision of information. On the other end, 25% was chosen for the current levels of reimbursement and information provision. These ratios were also suggested by the results of a questionnaire completed by a cohort of Dutch vascular surgeons (9).
Combining doctor and patient factors in a comparative model resulted in six hypothetical scenarios (80%-80%; 80%-25%; 50%-80%; 50%-25%; 30%-80% and 30%-25%). As both factors are interrelated (physicians will limit SET referrals if patients drop out due to reimbursement issues; conversely, physicians may not refer to SET due to presumed preference for an intervention), only the three most likely scenarios were considered: best (80%-80%), moderate (50%-80%) and worst case
(30%-25%). In addition, extrapolation for the Dutch population was performed by multiplying total savings on the total group by a factor 4.94 (21% of the Dutch population is insured by CZ).
The insurance database was analyzed with SAS (SAS Institute Inc, New York, USA). Differences between categorical variables were analyzed with a chi-square test. Statistical analyses were performed using SPSS 20 software (SPSS Inc, Chicago, USA). A two-sided p<.05 was considered statistically significant. Calculations of costs were calculated with Excel 2011 (Microsoft Office, Redmond, USA). Graphs were created with Graphpad Prism 5 (GraphPad Software Inc, La Jolla, USA).