Italian healthcare resource consumption for patients on hemodialysis treated for chronic kidney disease-associated pruritus (CKD-aP)

Data source

This study originates from the cross-linkage of the administrative healthcare data routinely collected in the ReS database, under specific agreements with several Italian local and regional health authorities (HAs), and periodically conveyed to the Italian Ministry of Health for reimbursement purposes. The INHS is a universal coverage single-payer healthcare system, thus the healthcare data collected by HAs represent the healthcare of all the INHS beneficiaries. Given the reliable representativeness of the Italian population (large catchment community and superimposable age distributions with those reported for the entire country by the Italian Institute of Statistics—ISTAT 1 (10)), Fondazione ReS has been conducting several observational studies on a range of clinical questions (11-13) since 2018. Its aims are to integrate findings from clinical trials and registries, which are still the most common bases of guidelines and recommendations, and to contribute to the evidence-based and patient-centered medicine. The following administrative databases establish the ReS database. The demographic database considers age, sex, HA of residency and disease waiver claim for co-payment. The pharmaceutical dataset consists of all drugs reimbursed by the INHS and supplied by local and hospital pharmacies: active substances can be analyzed by marketing code, Anatomical Therapeutic Chemical code (World Health Organization’s ATC classification) (14), dose, package number and dispensing date. The hospitalization database is analyzable through in-hospital diagnoses and procedures recorded in the hospital discharge forms related to overnight and daily hospitalizations (Italian version of the 9th International Classification of Disease—Clinical Modification [ICD-9-CM]) (15). The outpatient specialist care dataset is composed of examinations, diagnostics and invasive/non-invasive procedures performed in local facilities affiliated with the INHS and are analyzed based on the related current national classification system (2017 version of the “Nomenclatore tariffario”). Italian administrative healthcare databases also include all costs directly paid by the INHS, because of reimbursement purposes. The ReS database is physically placed into Cineca’s servers (16), whose collaboration guarantees compliance with international standard certifications of data quality and security. This retrospective observational study has analyzed Italian administrative healthcare data that are anonymized at the source and in aggregated form, according to the specific agreements with the HAs, owners of the data and European privacy laws. For these reasons and the institutional purposes of this study, neither informed consent nor ethical approval were applicable.

Cohort selection

From the ReS database, during the 3-year accrual period (2015-2017), patients with at least one in-hospital or outpatient hemodialysis procedure (ICD-9-CM code 39.95) were identified. About 5 million inhabitants captured annually by the ReS database correspond to about 15% of the Italian residents (10). The most recent date of hemodialysis represented the index date (supplementary figure 1). After having selected only patients with at least 2 years of ongoing hemodialysis before the index date, those potentially affected since 2013 by other diseases that generally cause pruritus (chronic liver disease/cirrhosis, systemic lupus erythematosus, arterial vasculitis and disease/inflammatory conditions of the skin and subcutaneous tissue) were excluded (supplementary table 1). Given the lack of a specific administrative code for CKD-aP and of most clinical information (e.g., dialysis registries, general practitioner [GP]’s database, laboratory values), the reimbursed supplies of gabapentin, pregabalin, thalidomide and antihistamines, recommended by the current guidelines (6,7), and the performance of UV phototherapy were used as proxies of CKD-aP. The final sample was categorized according to the presence/absence of the aforementioned CKD-aP-related treatments 180 days before and/or after the index date: gabapentin, pregabalin, thalidomide and antihistamines and the UV phototherapy (supplementary table 2). The time span of 180 days before and/or after the index date was chosen for the highest possible accuracy when associating CKD-aP and specific treatment supplies. CKD-aP-related therapies and the proxy that associates them with people on hemodialysis were suggested by the common clinical practice, the international guidelines (6,7) and the nephrologists who supported the researchers of Fondazione ReS for this study.

Epidemiological and clinical characterization

Age and sex at baseline were given. Moreover, within 2 previous years (until 2013), potential CKD-aP-related comorbidities (anemia, xerosis cutis, hyperparathyroidism, hypercalcemia, hyperuricemia and hyperphosphatemia) and other comorbidities of interest (hypothyroidism, diabetes mellitus, depression, arterial hypertension, viral hepatopathies and coronary artery disease [CAD]) were assessed (supplementary table 3).

CKD-aP-related treatments

The CKD-aP-related treatments reimbursed by the INHS were searched among pharmaceuticals (gabapentin, pregabalin, thalidomide and antihistamines), hospital discharges and outpatient specialist care (UV phototherapy) databases. Data 1 year before and 1 year after the index date were available for each patient (supplementary figure 1). The use of CKD-aP-related therapies was described within 1 year before and after the index date (to be consistent with the further follow-up analyses). DDD (defined daily dose—the assumed average maintenance daily dose of a drug used in its main indication (14)), drug packages’ and phototherapy procedures’ number, and patients supplied with a CKD-aP-related treatment at least once during one previous or subsequent year assessed the specific treatment consumption.

Healthcare integrated costs

One-year healthcare costs were assessed for CKD-aP and non-CKD-aP patients. Moreover, an individual matched pair case-control analysis was performed to assess the average annual cost by healthcare administrative database and overall, through an even more realistic perspective. Matched variables were sex, age and local HA of residency. Cases were people on hemodialysis potentially affected by CKD-aP, while controls were those without CKD-aP, according to the categorization by the presence or absence of CKD-aP-related treatments. Mean costs for cases and controls were compared through a z-test and a p-value <0.05 was considered statistically significant. Only INHS direct costs due to reimbursed pharmaceuticals, hospitalizations and outpatient specialist services are recorded in the Italian administrative databases. Specifically, pharmaceutical costs, by pharmaceutical group/active substance of interest, were calculated through sum and provided as mean per capita, starting from prices of community and hospital pharmacies (inclusive of value-added tax). The in-hospital expenditure was derived by the DRG (diagnosis-related group) system fee, which is used to calculate the reimbursed in-hospital stay costs per patient. Each DRG code corresponds to all in-hospital cares (from admission to discharge) in their entirety and complexity, without distinguishing single performed services. Local outpatient diagnostics and invasive/non-invasive procedures were assessed through the current national fees, listed in the 2017 version of the “Nomenclatore tariffario.” Pharmaceutical expenses were split into CKD-aP-related treatments and “other drugs” (i.e. all drugs different from the CKD-aP-related ones).

Statistical analyses

To compare patients’ demographics and comorbidities, a chi-square test between frequencies was performed, except for differences between patient distributions by age group, for which a chi-square test between distributions was performed. A p-value <0.05 was considered statistically significant. An individual matched pair case-control analysis was performed to assess the average annual cost by healthcare administrative database and overall, through an even more realistic perspective. Matched variables were sex, age and local HA of residency. The categorization of cases and controls was based by the presence and absence of CKD-aP-related treatments, respectively. Mean total costs per case and per control were compared through a Mann-Whitney non-parametric U-test, and a p-value <0.05 was considered statistically significant. All analyses were performed by means of Oracle SQL Developer Italian version 18.1.0.095 (California, United States).

Epidemiological and clinical characterization

From the 2015-2017 ReS population (on average about 5 million inhabitants per year), 6,147 patients (about 1.2 × 1,000/year) were treated at least once with in-hospital or local outpatient hemodialysis (Fig. 1), of which, 1,589 patients (1,589/6,147; 25.9%; 0.3 × 1,000/year) were treated for at least 2 years.

After having applied the exclusion criteria, the final sample (n = 1,239) was split into people with CKD-aP (n = 218) and without CKD-aP (n = 1,021). Both cohorts were mostly males and elderly (patients with CKD-aP were slightly older). Age distributions showed prevalence increasing with age, with a peak at 70-79 years (Tab. I).

The analysis of potentially CKD-aP-related comorbidities showed that hyperphosphatemia, hyperparathyroidism, anemia and hyperuricemia affected both cohorts in the same descending order of frequency (Tab. I). People on hemodialysis with CKD-aP were slightly more affected by all of them, except hyperuricemia. Among the other comorbidities, hypertension, diabetes mellitus, hypothyroidism, depression, CAD and viral hepatopathies were in the same descending order of frequency in both cohorts (Tab. I).

CKD-aP-related treatments

Table II shows that 58.3% (127/218) of people on hemodialysis with CKD-aP in the previous year and 65.1% (142/218) in the subsequent year received at least one CKD-aP-related drug. In both observation periods, about 10% of the cohort (23/218; 20/218) was treated with gabapentin and around 50% (108/218; 122/218) with antihistamines (53/218; 24.3% cetirizine), whose mean consumption (4.6 packs/patient) was in line with an annual chronic use in Italy. For all CKD-aP-related drugs, the annual mean consumption (package number and DDD) was very similar at one preceding and follow-up year. At least one phototherapy treatment/cycle was performed on 1.4% (3/218) of people on hemodialysis with CKD-aP in both periods and, on average, two times and four times per patient within one previous and subsequent year, respectively.

Healthcare integrated costs

During the follow-up year, on average, the INHS spent €37,065 per case, €31,286 per control and €35,988 per non-CKD-aP subject (Tab. III).

For all cohorts, the mean costs related to each administrative healthcare database similarly weighed on the overall expenditure: about 80% (€29,809/€37,065; €25,437/€31,286; €28,917/€35,988; respectively) was due to outpatient specialist services, about 10% to pharmaceuticals (€3,667/€37,065; €2,721/€31,286; €3,430/€35,988) and to hospitalizations (€3,590/€37,065; €3,129/€31,286; €3,640/€35,988). Most of the pharmaceutical expenditure was due to concomitant drugs. The hemodialysis performed out-of-hospital accounted for around 80% of the overall local outpatient care spending (€23,239/€29,809; €19,926/€25,437; €22,697/€28,917), while the in-hospital hemodialysis contributed more than 60% to the total expense for hospitalizations (€2,238/€3,590; €1,980/€3,129; €2,413/€3,640). High-efficiency dialytic therapies performed to people on hemodialysis with CKD-aP largely weighed on the overall mean annual cost. For each cost item and total healthcare expenditure, numerical but non-significant differences (p>0.05) between cases and controls were found.

Strengths and limitations

Administrative data represent a large and unselected community reliably reflecting the real population, with high accuracy in identifying patients with CKD (20). Nevertheless, limitations on the exclusive use of administrative data are several. First of all, given that a specific therapy for CKD-aP does not exist and the CKD-aP diagnosis is not available, a proxy had to be used for the identification of the cohort (i.e., supply of recommended therapies during 180 days before and/or after the index date + exclusion of some conditions frequently causing pruritus and treated by the same therapies (6,7)). Also, the inability to properly identify all the recommended therapies (e.g., therapies for mineral and bone disorders and topical treatments are not evaluable through administrative data (7)) and the INHS reimbursement conditions of antihistamines and gabapentin limit the findings. Indeed, the prolonged dispensation of antihistamines is reimbursed whether a chronic severe disease (e.g., the excluded ones) exists or in case of long-term treatment of seasonal allergic rhino-conjunctivitis (21), which is a condition hard to identify, but likewise uncommon; whereas gabapentin reimbursement is limited to specific neuropathies not including CKD-aP (22) and to epilepsy, leading to a misclassification.

The absence of out-of-pocket purchase, clinical information (e.g., dialysis vintage or dose, renal function) and other relevant patient characteristics could have contributed to slightly underestimate the cohort selection, also making it impossible to carry out some specific analyses. The ReS database still does not link to other databases, such as dialysis registries, GP’s operating system or those collecting laboratory outcomes, and consequently, some patient on hemodialysis with CKD-aP that is assisted by GPs, together with those whose healthcare is not reimbursed by the INHS, remain undetected by this study. Moreover, given the absence of many clinical variables and diagnoses mandatory for the prescription and for reimbursement purposes, only in-hospital diagnoses can be used for selection and analyses. This probably led to the inclusion of patients with an excluded disease on the one hand, and to the underestimation of people with comorbidities on the other hand. Moreover, the absence of a full panel of clinical variables has as limitation that a complete propensity score matching could not been performed, therefore we choose an exact/direct matching method 1:1 based only on the available and most reliable variables, such as sex, age and local HA of residency.

Finally, costs are slightly underestimated, mostly because they do not register the out-of-pocket purchase of healthcare services (e.g., over-the-counter drugs for CKD-aP seem very frequent (3)) and all indirect costs (e.g., those due to productivity loss or caregiver support).

The CKD-aP symptom burden is high and needs to be reduced early. The most updated guidelines (6,7) do not recommend a clear management, but recent studies and reviews (2,18,23,24) reported the tendency of a stepwise approach. Given the lack of an established and effective cure, the identification of comorbidities is essential to recognize the potential underlying mechanism and establish an appropriate therapeutic strategy.

The cost analysis is crucial for understanding the care pathway of people on chronic hemodialysis with CKD-aP and comparing it to that of patients on hemodialysis without pruritus. Our findings prove that the wide use of high-efficiency hemodialytic strategies was too expensive and should be delayed in favor of approaches based on strongest evidence, also following the discovery of an effective treatment (promising novel therapies, e.g., difelikefalin, are currently approved by the Food and Drug Administration [FDA] and European Medicines Agency [EMA] (24)). People on hemodialysis with CKD-aP still have significant unmet needs. Healthcare policy makers and payers should take action soliciting researches to improve treatments, but this effort accomplishes above all through the awareness of the healthcare professionals and patients themselves. Indeed, given the poor understanding of the pathophysiology of CKD-aP and lack of knowledge of its long-term outcomes (9), updated guidelines and education initiatives for clinicians and patients are deserved. Larger and higher-quality trials are needed (19), as well.