Healthcare utilisation, follow‐up of guidelines and practice variation on rhinosinusitis in adults: A healthcare reimbursement claims study in The Netherlands

Abstract Objectives To provide insight into healthcare utilisation of rhinosinusitis, compare data with clinical practice guideline recommendations and assess practice variation. Design Anonymised data from claims reimbursement registries of healthcare insurers were analysed, from 1 January 2016 until 31 December 2016. Setting Secondary and tertiary care in the Netherlands. Participants Patients ≥18 years with diagnostic code “sinusitis.” Main outcome measures Healthcare utilisation (prevalence, co‐morbidity, diagnostic testing, surgery), costs, comparison with guideline recommendation, practice variation. Results We identified 56 825 patients, prevalence was 0.4%. Costs were € 45 979 554—that is 0.2% of total hospital‐related care costs (€21 831.3 × 106). Most patients were <75 years, with a slight female preponderance. 29% had comorbidities (usually COPD/asthma). 9% underwent skin prick testing, 61% nasal endoscopy, 2% X‐ray and 51% CT. Surgery rate was 16%, mostly in daycare. Nearly, all surgical procedures were performed endonasally and concerned the maxillary and/or ethmoid sinus. Seven recommendations (25%) could be (partially) compared to the distribution of claims data. Except for endoscopy, healthcare utilisation patterns were in line with guideline recommendations. We compared results for three geographical regions and found generally corresponding rates of diagnostic testing and surgery. Conclusion Prevalence was lower than reported previously. Within the boundaries of guideline recommendations, we encountered acceptable variation in healthcare utilisation in Dutch hospitals. Health reimbursement claims data can provide insight into healthcare utilisation, but they do not allow evaluation of the quality and outcomes of care, and therefore, results should be interpreted with caution.


| INTRODUC TI ON
Rhinosinusitis is defined as symptomatic inflammation of the nasal cavity and paranasal sinuses. Acute rhinosinusitis (ARS) lasts <4 weeks, chronic rhinosinusitis (CRS) >12 weeks, patients with four or more episodes of ARS per year without symptoms in between are classified as recurrent acute rhinosinusitis (RARS). Patients with a prolonged or complicated course (eg meningitis, brain abscess, orbital cellulitis and orbital abscess) can also be distinguished. 1,2 Two or more symptoms should be present, one of which should be either rhinorrhea in ARS and RARS (anterior/posterior or both) or nasal blockage in CRS. Facial pain-pressure-fullness or loss of smell can also be present. For ARS and RARS, the diagnosis is confirmed by symptoms, for CRS by signs of inflammation at anterior/posterior rhinoscopy and/or pathological findings on CT. 1,2 There is a distinction between CRS with nasal polyps and without nasal polyps. 1,2 Worldwide, RS is a common disease, with a reported incidence of around 12%. 1 Of RS, ARS is most common and patients usually present themselves in primary care, at their general practitioner. 1 For CRS, a prevalence of 2% (defined with ICD-9 codes for primary care and referral centres) to 11% (defined by self-reporting) is reported, although there is a deficit in studies describing the prevalence of CRS in European countries. [2][3][4] RARS is less common, with a reported incidence of 0.03%. 5 Multiple clinical practice guidelines (CPGs) have been developed to guide and support clinical practice for RS, reduce practice variation and ultimately lead to better treatment outcomes. 6 In the Netherlands, a national CPG is available (CBO 2010), providing recommendations only on CRS. 7 Previous research shows that this CPG is used by most Dutch otolaryngologists. 8,9 In 2010, 61% of them reported being familiar with the CBO CPG. 8 More recently, research showed that 96% of Dutch otolaryngologists are aware of this CPG, with sufficient to good adherence to its recommendations. 9 However, data concerning actual CPG compliance in daily practice are lacking. 9 Despite CPGs that drive healthcare utilisation patterns, local or regional practice variations may exist, or systematic deviation from the CPG may occur.
In this study, we will use data from healthcare reimbursement claims registries of Dutch healthcare insurers to provide insight into the volume and cost of the RS-related healthcare utilisation in Dutch hospitals. We will compare results between different hospital types and regions to detect practice variation. In a previous study, healthcare reimbursement claims data have been used to assess non-adherence to guideline recommendations; therefore, we will compare our data to Dutch recommendations from the CBO 20120 guideline on CRS to detect potential deviations from protocol. 7,10 2 | ME THODS

| Ethical considerations
Under Dutch Law for Medical Research with Humans, it is allowed to process personal data for statistical and scientific analysis, provided data are not traceable back to individuals. 11 Data were provided, processed and analysed by Vektis, which is the national business intelligence centre of the Dutch healthcare insurers. 12 Data safety and security were guaranteed by Vektis. As we had no access to individual patient data, patient anonymity is guaranteed.

| Data extraction
We obtained data from Vektis, which collects and analyses healthcare reimbursement claims from almost all Dutch healthcare insurers, with coverage of >99% of healthcare providers. 12 The reimbursement procedure for healthcare insurance is the same across Data on age, gender and co-morbidity were obtained. To identify patients with comorbidities, we used a nationwide registration system on the use of pharmaceuticals (FKG). 15 Insured persons with a chronic condition were identified based on reimbursement claims of certain medication that is known to be used in a chronic condition. We extracted data on chronic obstructive pulmonary disease (COPD), asthma, diabetes mellitus and cardiac conditions since these are comorbidities known to influence decisions on surgical treatment strategy.

Key points
• We assessed healthcare utilisation for rhinosinusitis in the Netherlands, based on reimbursement claims registries for healthcare insurers.
• We obtained data on >99% of healthcare providers for 2016.
• Based on our data, there seems to be no structural deviation from most important clinical practice guideline recommendations.
• We found limited and acceptable practice variation.
• Healthcare reimbursement claims contain insufficient information to determine healthcare quality and treatment outcomes.
We obtained data on nasal endoscopy, allergy testing (skin prick), radiographic imaging (CT and X-ray), the number and type of surgical procedures and related hospital admissions.

| Comparison with Dutch CPG
We extracted 28 recommendations on the diagnosis and treatment of CRS for adults (see Supporting Information S1) and compared these to our data.

| Practice variation and comparison between hospitals
We compared practice patterns between different hospital types in the Netherlands and between three regions; South, North/East and West. (see Table 1). 16 North/East and South both have large rural areas, whereas West is more urbanised and densely populated. (see Figure 1).

| Data analysis
Performed in Microsoft ExcEl 2010. 17 Due to the large number of patients, 95% and even 99% confidence intervals are narrow, which results in differences of 1% already being statistically significant. 18 Since this is principally a descriptive study of the volume and the costs of care for RS in Dutch Hospitals, we will neither go beyond presenting the data distributions nor provide data on test statistics from statistical analyses.

| RE SULTS
We found a total of 56 852 patients with RS, that is a prevalence of 0.4% (total population ≥18 years of 13 585 073). 19 This accounts for 8% of patients that visited an otolaryngologist in the year 2016. 20 Costs of RS were €45 979 554-which is 0.2% of total hospital-related health care (€21 831.3 × 10 6 ). 21 Characteristics of patients can be found in Table 1. There was a slight female preponderance. Most patients were below 75 years old. Patients without co-morbidity were substantially younger than patients with comorbidity (mean age, respectively, 49 vs 61 years of age). COPD and/or asthma was more common than diabetes and/or cardiac conditions.
On average, patients visited the outpatient clinic 1, 3 times. 84% of patients visited the outpatient clinic more than one time, while 44% had three or more visits. For results of diagnostic testing, see Table 2. Details of surgical vs non-surgical patients can be found in Table 3. Surgery claims can be found in Table 4. Surgery was usually limited to the maxillary and ethmoid sinus; the frontal and sphenoid sinus were rarely operated. External sinus surgery was performed in <1% of cases. It was not possible to differentiate between solitary procedures, combined procedures (eg maxillary sinus and ethmoid surgery) and revision surgery. The majority of procedures were performed within daycare (see Table 5).

| Comparison with Dutch CRS guideline
Seven recommendations (25%) could be (partially) compared to the distribution of claims data 7 . See Figure 2.

| Healthcare utilisation compared by hospital and region
Patient characteristics can be found in Table 1. Most patients were treated in general/teaching hospitals and in the Western region, which reflects population density and hospital distribution in the Netherlands.(see Figure 1) Prevalence was similar across regions, with comparable patient population (based on age and co-morbidity), while costs per patient were lower in the denser Western region.

| Diagnostic testing compared by hospital and region
We found no important differences in claims for allergy testing by hospital type or region.
Nasal endoscopy was claimed most often in academic hospitals (75%) and private clinics (65%), and the least in general/teaching hospitals (60%). Nasal endoscopy was claimed somewhat more frequently in North/East and South (65% and 61%) compared to West (57%). CT scanning was claimed least in academic hospitals (27%) and was similar for other hospital types (49%). Claims for CT scans slightly varied across regions: North/East 44%, West 49% and South 52%. (also see Supporting Information S1).

| Surgical procedures compared by hospital and region
Surgical procedures were claimed least by private clinics (12%) and more often in other hospital types (teaching 17%, general 16%, academic 14%). Co-morbidity was present in 37% of operated patients in academic hospitals, vs 20% in private clinics. Surgery rates varied slightly between regions (West 15%, North/East 17% and South 18%), while co-morbidity of operated patients did not differ. In academic hospitals, relatively many external, sphenoid and frontal sinus surgeries were performed (24%, vs 0.5%-6% in other hospitals) and antral lavage was performed more often (7% vs 1%-3% in other hospitals). There were no differences in type of surgical claims between regions. Except for academic hospitals, the majority of surgeries was performed within daycare. In the Western region, 77% of surgery was performed within daycare, whereas in the North/East and South this was, respectively, 53% and 61%.

| Synopsis of key findings
We set out to assess healthcare utilisation and costs of RS for patients ≥18 years in secondary and tertiary care, based on healthcare reimbursement claims data, with a coverage of >99% of all healthcare providers. We discovered a lower prevalence than expected from previous studies. [1][2][3][4] Our study population was overall relatively young and healthy, which is comparable to previous studies. [1][2][3][4] Costs were less than one per cent of all Dutch hospitalrelated health care. For 25% of the recommendations in the Dutch CPG on CRS, diagnostic and treatment patterns could be (partially) compared using these Data. 7 Except for endoscopy, healthcare utilisation patterns showed no structural deviation from CPG recommendations, which is corroborated by limited regional practice variation. 7 However, our study shows major limitations, and on top of that, reimbursement claims are based on financial parameters and therefore do not allow evaluation of the quality and outcomes of health care.

| Comparison with other studies
Previous studies reported a higher prevalence of RS, although with the use of different methods. Less stringent definitions were used, studies were performed in primary care, or relied on self-reported symptoms. [1][2][3][4] Age and gender distribution corresponded to previous studies. 2,22,23 Allergy testing was encountered less than expected based on literature, but data on RAST are missing and allergy testing is probably invoiced using the diagnostic code "allergic rhinitis" which was not included in our study. 2,7,13,14 Results on nasal endoscopy are consistent with a study performed in the United States (concerning community and academic practice). 24 Nasal endoscopy is registered by Dutch otolaryngologists themselves which might lead to under-registration due to limited time and lack of financial incentive. However, this only partially explains the low number of registered endoscopies.
Surgery rate was comparable to a study in the United States. 23 Previous studies showed much higher surgery rate variation, that is, in the United States up to three times, in Finland up to four times and in Canada up to two times higher. [25][26][27]

| Variation by hospital type and region
In the Netherlands, most patients visit general and teaching hospitals; private clinics are not very common. There are eight academic hospitals in the country; these have an important function as referral centres for other hospital types (tertiary care), and therefore perform more complex care, which is reflected by our results.
We found little variation in the geographical prevalence of RS, which was to be expected since the geographic area of the Netherlands is small. For diagnostic testing, we found acceptable differences ranging from 1% to 8%. For surgery, rate differences were even smaller, being 2% or less. We did find a remarkable variation in the number of patients treated within daycare, which can be explained by the fact that in some regions patients generally live further away from the hospital.

| Strength and Limitations
We had access to a large database that covered more than 99% of patients in the Netherland; therefore, we can present an almost complete overview of all RS-related care. Our study is the first to assess healthcare reimbursement claims data for RS in the Netherlands.
However, major limitations have to be addressed.
First, the data are derived from reimbursement claims, which are financial outcome measures. In addition, incorrect registration and patients with duration of complaints of 8 weeks or more. 28 Therefore, patients with ARS will rarely be referred unless RARS or a complication is suspected, which is known from previous literature to be very rare. 1

| Implications
The general public and especially patients visiting an otolaryngologist can benefit from the new insights of this study and be reassured by the fact that we found little structural deviation from CPG recommendations.
Otolaryngologists can use our results for a better understanding of RS-related health care and comorbid diseases. Also, it shows in what way reimbursement claims can be used to assess health care.
In the future, they should be aware that secondary use of reimbursement claims data might increase and therefore adequate registration remains important.
Outcomes of this study can help CPG authors and board members in designing new or improved methods for healthcare delivery and registration in RS, from which patients eventually will benefit.
For example, further embedding guideline adherence as an evaluation tool in quality assessments might increase the adoption of evidence-based CPGs. Methods to assign these quality benchmarks have been recently developed. 29 For healthcare insurers and policymakers, it is important to realise that healthcare reimbursement claims data are too limited to assess quality of care or evaluate treatment outcomes. Our results also indicate the effects of market forces used to decrease healthcare costs. In line with competition between hospitals is higher in the Western region, we found decreased costs per patient, further contributing to a decrease of the total healthcare budget. Also, the low prevalence of RS in secondary/tertiary care, compared to the higher prevalence in previous studies (situated at population level or in primary care), implies that most patients with RS are treated by their GP and not by an otolaryngologist. [1][2][3][4] This demonstrates a high level of cost-effectiveness of the Dutch healthcare system.
For researchers, these results add to the existing knowledge about RS and can be used as a foundation for formulating research priorities. Our study can also serve as an example for future studies on healthcare reimbursement claims.

CO N FLI C T O F I NTE R E S T
None to declare.

AUTH O R CO NTR I B UTI O N S
NK and GH involved in concept of study design; NK involved in data acquisition; NK, MA, GH and RS analysed/interpreted the data; NK drafted the manuscript; NK, MA, GH and RS involved in critical revision of the manuscript; NK, MA, GH, RS involved in final approval.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data supporting the findings of this study are available from the corresponding author upon reasonable request.