Background
Herpes zoster (HZ) is a neurocutaneous infection caused by reactivation of the varicella-zoster virus (human herpesvirus 3) from the sensory ganglia when cell-mediated immunity wanes below a critical level [
1]. In a review of data from nine European countries, the overall incidence rate in the general population was approximately 100–400 per 100,000 person-years in adults aged 50 years or younger. Thereafter, it increased notably, reaching 1000 per 100,000 person-years after age 80 years [
2]. The risk of HZ is increased in patients with cell-mediated immunodeficiency, including that caused by drugs (e.g., corticosteroids or chemotherapy), radiotherapy, viral infections (
i.e., human immunodeficiency virus [HIV]), or malignancy.
HZ may have severe acute complications, such as meningo-encephalitis and cutaneous dissemination, in particular among individuals with the aforementioned predisposing conditions [
3,
4]. These patients are more likely to be treated in the hospital or in hospital outpatient specialty clinics. When assessing the epidemiology of HZ, it is important also to consider these severe or potentially severe cases diagnosed in a hospital-based setting. Nevertheless, results from nationwide population-based hospital registries in Europe are sparse concerning the diagnosis rate and patient characteristics, including prevalence of complications and comorbidities. In particular, none of the previous population-based studies published data pertaining specifically to diagnoses from hospital outpatient specialty clinics. To address this paucity of data, we conducted a nationwide Danish cohort study describing HZ in a hospital-based setting, including diagnosis rate, patient demographics, severity, comorbidity burden, length of admission, and subsequent all-cause mortality.
Methods
The Danish National Health Service provides tax-funded universal health care to all residents (population 5.6 million), guaranteeing unfettered access to general practitioners and hospitals [
5]. Utilization of health services is recorded in nationwide databases using the unique central personal register (CPR) number assigned to each Danish resident at birth or immigration [
5]. The CPR number allowed for unambiguous individual-level linkage of the registries used in the current study.
The Danish National Patient Registry (DNPR) contains data on all admissions to Danish non-psychiatric departments since 1977 and visits to hospital-based outpatient clinics (
i.e., ambulatory medical care) and emergency rooms since 1994 [
6]. Each DNPR record contains the CPR number of the patient, dates of hospital admission and discharge or start and end of outpatient contact, information on procedures performed, and diagnoses. Each record includes a primary diagnosis,
i.e., the condition considered the main reason for the hospital contact. Diagnoses are assigned by the discharging physician at the time of hospital discharge or at the end of an outpatient contact using the
International Classification of Diseases, 8th revision (ICD-8) until the end of 1993 and 10th revision (ICD-10) thereafter [
6]. Surgical procedures are coded by the surgeon according to a Danish classification system (1977 through 1995) and a Danish version of the Nordic Medico-Statistical Committee Classification of Surgical Procedures (from 1996 onwards). Since 2002, the DNPR has provided the basis for reimbursing both public and private hospitals. Reporting of all services provided in the private hospital sector became mandatory in 2003. General practitioners and private practice specialists do not report to the DNPR. Diagnosis codes and definitions used are provided in Additional file
1: Table S1.
We used the DNPR to identify all first-time inpatient, outpatient, and emergency room diagnoses of HZ. To ensure complete information on outpatient and emergency room diagnoses and to avoid potential problems associated with the transition between the ICD-8 and ICD-10 coding systems, we chose a study period from January 1, 1994 to December 31, 2012, when ICD-codes were used, exclusively. The date of discharge or end of outpatient contact was considered the diagnosis date. Based on patients’ first-ever HZ diagnosis recorded in the DNPR, we classified HZ diagnoses according to type of hospital-based contact (inpatient, outpatient clinic, or emergency room) and type of diagnosis (primary or secondary). We also identified individual ICD-10 codes for HZ. However, because of the clinical difficulty of distinguishing between HZ encephalitis and meningitis, we combined ICD-10 codes B02.0 and B02.1 in our analysis. Similarly, we grouped ICD-10 codes according to severity (complicated or uncomplicated) and extent (disseminated or localized), as shown in Additional file
1: Table S1. The difference between groups of complicated and disseminated HZ is attributed to codes for HZ ophthalmicus, which is regarded as a localized but complicated form.
To give an overview of reasons for healthcare contacts among patients with a secondary HZ diagnosis, we categorized the primary diagnosis according to a list of 203 morbidity groups based on the ICD-8 and ICD-10 World Health Organization morbidity tables. This code list has been published previously [
7] and is outlined in detail in Additional file
1: Table S1. Additional codes outside this list comprised general and unspecific reasons for hospital contact. We grouped them into four categories (‘Symptoms, signs and abnormal clinical and laboratory findings not otherwise specified’, ‘Injury, poisoning and certain other consequences of external causes’, ‘Observation for or follow-up after treatment for cancer’, and ‘Contact with health services due to solid organ or bone marrow transplantation, and ‘Other factors influencing health status and contact with health services’). The HZ vaccine, which is approved by the European Medicines Agency for use in adults aged 50 years or older, did not become available in Denmark until September 2014, and hence not considered in the present study [
8].
To quantify comorbidity, we obtained information on all primary, secondary, and additional diagnoses recorded in the DNPR before or on the same date as the HZ diagnosis. We classified comorbidity using the Charlson Comorbidity Index (CCI), a scoring system that assigns between one and six points to 19 groups of chronic diseases (Additional file
1: Table S1) according to their ability to predict death [
9]. We obtained information on diagnoses in each disease group. Based on the total CCI score, we also categorized patients into levels of none (score 0), moderate (score 1), severe (score 2), and very severe (score 3 or more) comorbidity. We also obtained information on diagnosis history of conditions associated with immune dysregulation (
i.e., stem cell or bone marrow transplantation, solid organ transplantation, HIV infection, primary immunodeficiency, any cancer, and autoimmune diseases) and thus potentially an increased HZ risk [
1]. For the categories HIV infection, any cancer, and autoimmune diseases, there was an overlap with codes included in the CCI.
The Danish Civil Registration System (CRS) provides a record of changes in vital and migration status for the entire population since 1968 [
5]. We used the CRS to follow HZ patients from diagnosis until the date of death, emigration, or December 31, 2012.
Statistical analysis
We computed the annual rate of hospital-based HZ diagnoses throughout the study period, with direct age standardization [
10] to the 2000 Danish census. We characterized patients according to age, sex, comorbidities, calendar period of diagnosis (1994–1999, 2000–2005, and 2006–2012), and types of HZ diagnoses. For patients with an inpatient primary HZ diagnosis, we computed the median and total length of hospital stay.
We used the Kaplan-Meier estimator to compute absolute mortality risks. We computed standardized mortality ratios (SMRs) with 95 % confidence intervals (CIs) for HZ patients using indirect standardization [
10], with the general Danish population as reference. The expected number of deaths was obtained by multiplying the number of person-years at risk in the HZ cohort with the rate of death in the general population according to 5-year age groups, sex, and calendar year. The SMR was computed as the ratio of the observed to the expected number of deaths. Corresponding 95 % CIs were computed assuming that the observed number of deaths followed a Poisson distribution. We performed analyses overall and in subgroups of demographic and patient characteristics.
Analyses were performed using STATA® (version 12.1, STATA, College Station, TX). The study was approved by the Danish Data Protection Agency (record number 1–16–02–1–08). Danish legislation does not require ethical review board approval or informed consent from subjects in registry-based studies.
Discussion
In this nationwide Danish study, the rate of hospital-based HZ diagnoses was stable at around 13 per 100,000 person-years over a study period of 19-years. The rate was 3–6 per 100,000 person-years for persons aged below 50 years, increasing by approximately twofold with each consecutive decade of age. Thirty percent of patients had complicated HZ and comorbidity was prevalent, particularly diseases involving immune dysregulation.
The age-specific hospitalization rates from the present study and previous European studies are shown in Additional file
1: Figure S3. While the overall and age-specific inpatient hospital diagnosis rates in Denmark accord with previous studies [
11‐
24], however there is considerable variation across countries. This variation may result from differences in hospital admission practices or health-seeking behavior, as the overall incidence rate in the general population varies only slightly across Europe [
2]. Nevertheless, the nearly exponential increase in the rate of hospital-based HZ diagnoses after age 50 years [
11‐
18,
20‐
23] and the female predominance [
12,
13,
16,
18,
19,
21] in our study are comparable with results from the previous studies. However, none of these studies presented data specifically on HZ diagnosed in hospital-based outpatient clinics, which accounted for one-fourth of patients in our study.
We observed a shorter length of hospital stay (mean 6.2 days) compared to previous studies (mean 7.5 to 11.2 days) [
12,
13,
17‐
19,
22,
25,
26]. Our result was closest to that reported in a Swedish study [
18], which also had a similar diagnosis rate and proportion of complicated infections. The shorter length of stay may be explained by hospitalization of milder cases of HZ in Denmark and Sweden, as the proportion of complicated infections (mainly ocular and neurological) was approximately twice as high in Southern Europe [
12,
14,
19,
20,
25,
26]. Studies from Southern Europe also reported a higher prevalence of predisposing conditions, such as HIV [
12,
14,
16,
20], organ transplantation [
12,
16], diabetes [
16], and ‘blood dyscrasias and some immune deficiencies’ [
16] in HZ patients. Frequencies of malignancy [
14,
16,
20] and connective tissue disease [
16] were comparable to the results in our study. Finally, the length of stay for patients with neurological complications in our data accords with Italian and French data [
19,
25,
26], further suggesting that differences in admission practices play a role. The slight temporal increase in the diagnosis rate of HZ encephalitis and meningitis in our study may result from increased use of polymerase chain reaction technology for detecting varicella-zoster virus in cerebrospinal fluid [
27].
Chronic pulmonary disease, connective tissue disease, cancer, HIV, and solid organ transplantation were relatively more frequent in our study compared with other selected Danish patient populations [
28‐
31]. This finding is consistent with the increased risk of HZ in patients with immune dysregulation due to either disease pathophysiology or treatment (e.g., with glucocorticoids) [
1]. Apart from posing an additional burden to these patients, HZ could complicate their treatment, for instance by delaying chemotherapy for cancer. We lacked data on HZ treated in general practice and had no comparison cohort, which prevented us from examining whether HZ was associated with increased mortality in patients with certain comorbidities. Although we found increased mortality among patients with comorbidity, we obtained this result using standardization to the entire general population, including those without comorbidity. Thus, the increased SMRs may potentially reflect excess mortality from the comorbid condition rather than from an interaction between HZ and the comorbid condition. The difference in SMRs by age could also be explained by use of the general population as reference. A high prevalence of the exposure (HZ) in the general population would bias the SMR towards the null [
32]. Because the prevalence of HZ increases with age, the risk of this bias also may increase with age, potentially explaining our finding of lower SMRs in the oldest age groups.
We used a comprehensive nationwide database to study hospital-based HZ diagnoses within a universal healthcare system, which limits selective inclusion of specific hospitals, insurance groups, or age groups. Nonetheless, our study has some limitations. Since 2002, reimbursements to public and private hospitals have been administered using the Diagnosis Related Group (DRG) system in the DNPR [
6], which may have affected physicians’ choice of primary and secondary diagnosis codes. As well, the completeness and validity of HZ diagnoses in the DNPR is unknown. Still, other studies have found high diagnostic accuracy with positive predictive values ranging from 72 to 100 % [
33‐
37]. Although there is a specific code for specifying consultation with post-herpetic neuralgia (ICD-10 G53.0), it is possible that some codes, particularly ‘other nervous system involvement’, were assigned alone to patients who had this chronic complication rather than acute HZ. However, we consider such misclassification of minor importance because these patients experienced recent varicella-zoster virus reactivation.
Our comorbidity assessment has some limitations. The positive predictive value for diagnoses included in the CCI are known to vary depending on the disease, department of origin, and reference standard [
6]. Moreover, incompleteness is expected for conditions that are treated primarily in general practice, e.g., diabetes. Another concern is underestimation early in the study period due to left-censoring of outpatient and emergency room diagnoses before 1994. Similarly, changes in coding when new medical treatments are introduced may explain the lack of HZ patients with bone marrow transplantations early in the study period.
The majority of patients with HZ in Denmark are treated in general practices [
38], which are not subject to systematic monitoring in national registries. Thus, analysis of temporal trends in our study may not reflect true changes in disease incidence in the population. Administrative factors (e.g., changing admission practices) are also important in determining whether patients receive a hospital-based diagnosis. As well, the lack of primary care data may have caused underestimation of the SMRs [
32].
Acknowledgement
This work was supported by ‘Dagmar Marshalls Fond’, ‘Friedrich Wilhelm Frank og Hustru Angelina Franks Mindelegat’, ‘Kgl. Hofbuntmager Aage Bangs Fond’, Aarhus University, and the Program for Clinical Research Infrastructure established by the Lundbeck and the Novo Nordisk Foundations. None of the funding sources had a role in the design, conduct, analysis, or reporting of the study. The Department of Clinical Epidemiology, however, is involved in studies with funding from various companies as research grants to (and administered by) Aarhus University. None of these studies have any relation to the present study.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SAJS conceived the study idea, carried out the analyses, reviewed the literature, and wrote the initial draft. JK performed initial data management. SAJS, JK, MV, HCS, and HTS participated in designing the study, critically revised the manuscript for intellectual content, and approved the final version.