Patterns of prescription, hospitalizations and costs of herpes zoster in patients at risk, from a large Italian claims database

Introduction

Herpes zoster (HZ), also known as ‘shingles’, is a common and debilitating disorder caused by reactivation of varicella zoster virus (VZV), which is typically contracted during childhood. Following the primary infection, which manifests clinically as chickenpox, the virus becomes latent in the sensory cranial nerve and spinal ganglia and may reactivate, giving rise to nervous and skin symptoms with a dermatomal distribution typical of HZ (1).

In Europe, it is estimated that more than 1.7 million people suffer from HZ. The estimated overall mean incidence is about 3.4-4.8 per 1,000 person-years. The risk of HZ episodes increases with age, up to more than 11 per 1,000 person-years among 80-year-old subjects (2). Several medical conditions may increase the risk of HZ occurrence or induce symptom worsening. Age and immunosuppression seem to be the major risk factors, but also other conditions, including diabetes mellitus, cardiovascular (CV) disease and chronic obstructive pulmonary disease (COPD), have been shown to be associated with an increased risk of HZ (3-7). The Italian National Vaccine Prevention Plan (PNPV) 2017-2019 presented the new timetable for the vaccinations recommended and reimbursed by the Italian National Health System (NHS), by age and risk groups. Among the novelties, the zoster vaccine has been introduced for people aged 65 and older, and for subjects ≥50 years with at least a disease that may increase the risk of HZ or cause symptom worsening (CV disease, COPD, diabetes and immunosuppression) (4). Immunization can reduce both HZ incidence and complication rates, including ocular and postherpetic neuralgia (PHN)-related pain (estimated in approximately 20% of cases of HZ) (4,8,9).

This observational study, based on real-world data, aimed at identifying the incidence of HZ in patients aged 50 and older with at least one disease that increases the risk of HZ, according to the last PNPV. Additional purposes of this study were to characterize the HZ incident subjects and to determine their annual healthcare resource utilization and costs from the NHS perspective.

Methods

Results

Within the ReS database, 12,562,609 Italian inhabitants had complete data of their healthcare services provided by the NHS from 1 January 2012 to 31 December 2015. The ReS database consists of administrative data accurately representative of the Italian population. Indeed, distributions by age and gender are consistent with those provided by the Italian Institute of Statistics for 2013 (Supplemental Figure 1, Percentage distribution of the 2013 Italian population by age group, from the ReS database and from the Italian Institute of Statistics (ISTAT), available online), and the results provided by this observational analysis reflect the real demographic characteristics of candidates to HZ vaccination.

Among patients aged 50 and older (n = 5,419,586), 1,004,705 (18.5%) had at least one risk factor for HZ during 2013 and were not hospitalized for HZ and/or were not prescribed brivudine over the previous year. In particular, 87.1%, 11.9%, 1.0% and 0.08% of subjects had one, two, three and four risk factors, respectively. Specifically, 58.9% of patients were affected by diabetes, 29.5% by CV disease, 22.7% by COPD and 2.8% by immunosuppression (Fig. 1). Certainly, a clinical assessment is necessary in order to confirm and integrate comorbidity cohorts (e.g. administrative databases do not identify patients with diabetes treated only with diet) as well as to define the severity of target clinical conditions and of related comorbidities. Over the 2-year follow-up, 5,916 patients of the HZ at-risk cohort experienced HZ, resulting in a 2-year incidence rate of 5.9 per 1,000. We are fully aware of the limitations of administrative data analyses and of the possibility of underestimating the incidence of HZ. That is why we wanted to make sure that our HZ incidence findings were as much aligned as possible with reality. Nevertheless, the calculated annual incidence was 2.9 per 1,000 person-years among individuals aged ≥50 years at risk for HZ. This is about a half of the 2-year rate; therefore, this is what we expected by considering a minimal loss to follow-up. Although this 1-year incidence rate is not directly comparable with findings from other studies, probably due to differences in both study designs and populations analysed, it appears similar to the estimates across the world (3.0-5.0 per 1,000 person-years), Europe (2.0-4.6 per 1,000 person-years) and Italy (1.59-6.9 per 1,000 persons) (4,7,14,15), recorded among patients ≥50 years with comorbidities (16).

Specifically, 92.1% of subjects were identified through brivudine prescriptions, 7.1% through hospitalizations for HZ (primary/secondary diagnosis) and 0.8% by both criteria. Since HZ is typically diagnosed and treated in the outpatient setting, hospitalizations represent the most severe cases. The lack of general practitioners’ (GPs) visits and their associated diagnoses in the ReS database could be considered a limitation of this study, but hospital prescriptions of brivudine are very few, and therefore this criterion could reliably identify the HZ cases diagnosed by GPs. Certainly, the HZ cases among outpatients could be underestimated as they could be treated with other antivirals (i.e. acyclovir, valacyclovir and famcyclovir) not included in the selection criteria. Indeed, they could not have been specific for HZ cases, because their reimbursement circumstances (Nota AIFA 84 (17)) also include some herpes simplex virus and VZV conditions.

The mean age (±standard deviation) of patients with a new HZ event was 74 ± 10 years and 54.3% were female (Tab. I). The 2-year incidence increased with age, reaching the highest values in the 80- to 89-year-old group (7.2 per 1,000) and among females (6.7 per 1,000), consistently with the literature (6,7,18). Within risk factor groups, the 2-year incidence rates were, in ascending order: 5.6/1,000 among diabetics, 6.1/1,000 in patients with CV disease, 6.8/1,000 in patients with COPD and 6.9/1,000 in patients with weakened immune system. In accordance with the literature, immunosuppressed patients experienced the highest incidence (6.9/1,000 vs. 5.6/1,000 in diabetics: p<0.01). This suggests that the vaccine could provide the support needed to the immune system to prevent the expression of HZ (6, 16). The forthcoming non–live recombinant subunit vaccine could fulfil this need, as to date it is not contraindicated in immunocompromised subjects. Moreover, although the 2-year incidence rate increased with the number of risk factors from 5.9/1,000 (n = 1) to 12.2/1,000 (n = 4), it was independent of them (p>0.05).

Specific healthcare consumption and costs were assessed for 5,455 patients (92.2% of the incident cohort) who could be followed up for 1 year.

In order to evaluate the recurrent use of antivirals after a new diagnosis of HZ, the 1-year rate of patients treated with at least one antiviral drug among brivudine, acyclovir, valacyclovir and famcyclovir was analysed (Tab. II), with 82.2% of patients who were not lost to 1-year follow-up. On average, each patient costed yearly €106 to the NHS (acyclovir and valacyclovir having the lowest cost) and received 1.4 boxes (from 1.1 of brivudine to 3.7 of acyclovir). According to risk factors, the mean annual cost per treated patient ranged from a minimum of €105 for diabetics to a maximum of €115 for immunosuppressed patients.

Literature data on medical outpatient care costs show discrepancies (14). By evaluating only the supply of antivirals, this study found a 1-year cost for specific medicines that accounted for 21.6% of the total expenditure for a patient experiencing HZ in the presence of risk factors, in the perspective of the NHS. This is in line with the proportion of patients treated with antiviral therapies found by a Dutch study, and with the costs reported in European studies ranging from €118 to €242 regardless of age (14,16).

Another specific analysis aimed to evaluate the use of recommended systemic pain therapies (pregabalin, gabapentin and tramadol) during the first 6 months after the HZ episode, identifying 1,435 patients treated with at least one pain drug (Tab. II). On average, each patient costed €94 to the NHS and received 4.6 boxes of pain drugs during the 6-month follow-up. Since there is poor consensus on the definition of postherpetic pain duration (2,6,15,16,18,19), which may occur until 1 month (acute pain), from 30 to 90 days (subacute pain) or from 3 to 12 months (PHN) from diagnosis, the choice to search for these pain therapies over a 6-month period after the new HZ event could have generated both an overestimation and an underestimation of the complications’ severity. Moreover, the use of pharmaceuticals rather than hospitalization codes for PHN could also have underestimated its burden.

This real-world data analysis allowed to assess only the expenses reimbursed by the NHS. An underestimation of all the direct costs of patients with HZ was inevitable, because administrative databases do not consider the private outpatient care. The last official report on the Italian pharmaceutical expenditure estimated that 28% of the mean per capita expenses for drugs in 2018 are paid by the patient (20). In addition, subjects who deny the use of specific drugs reimbursed by the NHS should also be taken into account. The limitations of the healthcare resource availability from the administrative databases can be an obstacle for decision makers to evaluate the real-world evidence (RWE) (21).

The 1-year rate of hospitalizations with a primary diagnosis of HZ, with or without complications (ICD-9-CM code 053.x) of incident patients not lost to follow-up (n = 5,455), was 8.0%. They were hospitalized for an average of 16.6 days and 1.1 times (Tab. II). The mean annual cost of a patient hospitalized due to HZ was €3,927. If risk factors are considered, immunosuppressed patients were admitted to hospital more frequently (16.8%) than the others, resulting also in the most expensive subjects, with an average cost of €5,172 per hospitalized patient. In contrast, patients with diabetes were less frequently hospitalized (6.0%) and patients with COPD had the lowest hospitalization costs (€3,550 per patient). In our analysis, the 1-year hospitalization rate due to HZ and the mean length of stay seem higher compared to those reported in other studies (16, 22). Nevertheless, although this analysis did not distinguish between hospitalizations for HZ and those for HZ-related complications (e.g. PHN, zoster ophthalmicus), this strategy was grounded on the fact that the use of the primary diagnosis with the ICD-9-CM code 053.x, including HZ and its complications, can be highly predictive of an episode of HZ, especially in the older adult population (22). Moreover, the longer length of stay can be due to the restriction of the study cohort to subjects with comorbidities, such as CV disease or COPD. Although there are some discrepancies in estimating the NHS absorption of the expenses of hospitalizations due to HZ and its complications, the annual mean inpatient care cost is consistent with that of Panatto et al (14).

In the additional investigation comparing the HZ at-risk cohort to controls (Tab. III), the difference in the 2-year incidence rates was statistically significant (5.9/1,000 vs. 4.7/1,000; p<.01), confirming this discrepancy also by gender and age.

On average, the NHS yearly spent €402 for a patient with HZ or at risk for HZ vs. €268 for a patient with HZ but without the risk factors considered.

Conclusions

Even if it is not yet included in the routine work for outcomes research of companies, some regulatory agencies (i.e. AIFA for Italy, and Food and Drug Administration for the U.S.A.) use RWE for healthcare decision-making. Differently from randomized clinical trials that are conducted in homogeneous populations, observational studies of real-world settings allow generalizability of the results to the actual population (23). Due to the large size and the representativeness of the Italian population of the ReS database, it was possible to reliably identify and characterize subjects for vaccination against HZ. Analysis of administrative databases can help healthcare stakeholders to study consistently the real burden on the NHS, or rather the costs potentially related to specific diseases and populations. Estimating the size of the burden of individuals eligible for the zoster vaccine will permit to define the impact of vaccination programmes, in terms of health benefits and cost savings. The goal of the PNPV 2017-2019 was to reach in 2019 a vaccine coverage against HZ of 50% among people aged ≥65 years (4). It has been estimated that if the zoster vaccine was extensively administered to the target population, there should be an important decline in the incidence, severity and duration of complications (e.g. ocular and PHN) (9,24). The PNPV recommends the zoster vaccine to immunocompetent people aged 65 and older and to subjects from the age of 50 with risk factors, including the immunocompromised state, without a recent history of HZ. Considering an incidence rate of 6.3 per 1,000 (2005) in people aged ≥50 years, the PNPV reported that in Italy the economic burden of HZ and PHN could be estimated at €41.2 million, including both direct (two-thirds of the total expenditure, about 28 million) and indirect costs (4). With the aim of reaching 50% of vaccine coverage in 2018, the saving has been estimated at €13,868,063. This estimation will tend to increase every year because the number of new patients vaccinated will be added to patients already immunized against the zoster virus. Therefore, also the savings for the State will increase. This real-world observational analysis provided an annual HZ incidence of 2.9 per 1,000 person-years in subjects aged 50 and older with clinical conditions considered as risk factors, providing a measure of the burden of this condition in the perspective of the NHS. Therefore, our findings can help the health governance to improve clinical decisions and economic positioning concerning the zoster vaccine plan.

Acknowledgements

The authors thank Paola Luciolli and Luca De Fiore (Il Pensiero Scientifico Ed.) for the professional revision of this article.

Disclosures

Conflict of interest: The authors have nothing to disclose.

Financial support: This research received the unconditional support of GlaxoSmithKline Biologicals SA (study number: 208681), which has not been involved in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.