Increasing incidence associated with herpes zoster infection in British Columbia, Canada

Varicella zoster virus (VZV) infection or chickenpox typically occurs in children and manifests as a diffuse itchy rash. The rates of infection are high with approximately 95 % of the population testing seropositive for VZV [1, 2]. Reactivation of the varicella virus later in life is known as herpes zoster (HZ) infection [3], and is triggered when an individual’s immune system, cellular immunity in particular, decreases due to the process of aging [4] or becomes immunocompromised [5]. Other risk factors for development of HZ include race (Caucasians are at higher risk than African Americans), gender (most studies show a higher incidence among women), stress, trauma, and diabetes [6‐8]. Herpes zoster infection manifests itself as a unilateral, vesicular, painful rash [9], and can lead to complications such as post-herpetic neuralgia (PHN) [10, 11] in approximately 20 % of people.

Recent studies from Europe [12] and North America [13] have shown a trend of increased zoster infection rates over time. Scientists are unclear as to the reason for its increase, but Hope-Simpson in his original studies had hypothesized that viral latency after infection with VZV is maintained by “immunosurveillance”, which is boosted by periodic subclinical reactivations and exposure to exogenous virus [3, 14, 15]. Thus, HZ clinically manifests itself when “immunosurveillance” falls below a certain threshold [3]. Some have hypothesized that the introduction of widespread childhood varicella vaccination programs could lead to less wild type virus boosting for adults [16‐18], leading to an increase in zoster rates. Alternatively, the aging population and presence of larger numbers of immunocompromised individuals (due to autoimmune diseases, transplants and larger numbers on medications such as corticosteroids, DMARDS and biologics) [19] could also account for the increase in zoster rates seen around the world [12, 13].

Given that it has been a decade since Canada evaluated its rates of zoster infection [20, 21], we undertook this study to determine the trends over time in age and sex-specific herpes zoster and PHN rates. We further examined the rates of hospitalization and costs associated with HZ and PHN over time. Finally, although not our primary objective, we looked at the impact of varicella vaccination on herpes zoster rates.

In our study, over the 16 year period, we saw a 1.5-fold increase in HZ incidence regardless of the definitions used to identify herpes zoster. For young children between the ages of 0 and 9 years, the HZ incidence decreased abruptly from 1.8 to 0.8 per 1000 population in 2004, and has been steadily declining, likely reflecting the start of the publicly-funded varicella vaccination program which targeted children in kindergarten (age 6 years). Interestingly, despite the introduction of the varicella vaccination program in 2004 for 12 year olds, the age-specific HZ incidence held steady but did not decline for those within the 10–19 year old category. This is probably related to the fact that most of them will have had varicella infection already and have good immunity; for now too little time has elapsed to show a decrease in this age groups of HZ rates. Alternatively, the lack of effect on HZ rates may be related to a lower vaccine uptake rate in this age group than the kindergarteners or the fact that the vaccine effectiveness was reduced with the one dose program. On the positive side, zoster rates have not increased either, as they have for all other age categories after the age of 19 years. Although we saw a slight increase in zoster incidence in the 20–39 year old age groups, the largest increases were seen in the 40–49 (63 %) and the 60–69 year olds (61 %). The 80 year old group had the highest incidence in 2012 (11/1000 population), although their increase was only 21 % between 1997 and 2012. Our finding that HZ incidence increases with age has been reported in other studies and is thought to be related to immunosenescence, that is, decreasing ability to respond to the reactivated virus as the body ages [4, 19]. Pinchinat and colleagues [12] in their review of studies looking at the incidence of HZ across Europe showed that HZ incidence increases sharply with age, from around 1/1 000 children <10 years up to 10/1 000 people over 80 years of age. Similarly Kawai et al. found that age-specific incidence was similar across countries and rose sharply after the age of 50 years; their rates were similar to our study as they found rates of 6-8/1000 person-years at age 60 years and 8-12/1000 person-years at age 80 years [13].

Our observation of a higher zoster incidence in women has also been seen in other studies [7, 32‐34]. Originally, investigators believed this difference was not a real phenomenon and it was likely due to women having greater health-seeking behavior. However, given that the difference between females and males occurs in all age groups, including children, and has been seen consistently in many studies using different methodologies, it is unlikely that this phenomenon is due to differential health-seeking behaviours. The biological reason behind this effect is unclear; it may be related to symptomatic zoster being more common in women, or gender differences in immune responses to the varicella antigen and loss of immunity over time [7, 35, 36].

In our study, the standardized rates seen in 2012 for zoster were within the range reported in other North American and European studies [12, 13]. In their systematic review Kawai et al. included 130 studies conducted in 26 countries. They found the incidence of HZ ranged from 3 to 5 per 1000 person-years [13]. Although the methodology used in these studies varied from prospective surveillance to use of retrospective administrative data to determine incidence, most of the studies showed a temporal increase in the incidence of HZ and PHN in the past several decades. The authors also observed that those studies which evaluated all age groups (rather than restricting the analyses to those over 60 years of age) and were conducted in the 50s to early 90s, showed lower HZ rates (HZ incidence of North American studies ranged 1.31 to 2.40 per 1000) [37, 38] than studies conducted 2000s (HZ incidence of North American studies ranged 3.82 to 5.79 per 1000) [35, 36, 39‐41].

In our secondary analysis we evaluated the impact of the introduction of the publicly funded varicella vaccination program. Between 2007 and 2012, the average coverage rates in the 2 year old, kindergarden (6 year old) and grade 6 (12 year old) children was 83 %, 74 %, and 63 %, respectively [42]. Despite these uptake rates, a vaccine effect was not seen after adjusting for potential confounders which would have an effect on HZ incidence, including age, sex, and immunosuppression status. Previous older studies using survey data and modeling had suggested that the introduction of a routine varicella vaccination program could potentially increase the incidence of zoster infections due to less boosting of immunity by the wild type virus [16, 18, 19, 43, 44]. However, North American studies conducted in the last 5 years, using large databases have shown that although age-specific HZ incidence is increasing, it started increasing before the introduction of widespread varicella vaccination. Hales et al. [45] conducted a retrospective study using Medicare claims to examine the link between herpes zoster incidence in the US population older than 65 years and childhood varicella vaccination. They found that age and sex standardized HZ incidence increased 39 % from 10.0 per 1000 person-years in 1992 to 13.9 per 1000 person-years in 2010. They used Poisson regression analysis to compare HZ trends during 3 periods of varicella vaccination program implementation: preimplementation (1992 to 1995), early implementation (1996 to 1999), and full implementation (2000 to 2010). The authors also found that the rise in HZ incidence predated 1996, when the U.S. Advisory Committee on Immunization Practices first recommended routine varicella vaccination for children aged 12 to 18 months, and HZ incidence did not accelerate after full implementation of the varicella vaccination program, when vaccine coverage reached 90 % and varicella incidence decreased by 90 %. The authors of this paper also looked at HZ incidence in US states that had low varicella vaccine coverage rates compared to high coverage rates; they found that state varicella vaccination coverage had no effect on HZ incidence concurrently or 10 years later (RR, 0.9998 [CI, 0.9993 to 1.0003]) after adjusting their model for sex, age, and calendar year [44]. Using medical claims data, Leung and colleagues [38] evaluated HZ incidence in all persons enrolled in the MarketScan® databases (Thomas Reuters, Ann Arbor, MI). They used similar definitions as our study and defined incident HZ as an enrollee of any age in the MarketScan® database with an outpatient claim bearing a HZ ICD-9 code (053.xx) in the primary or secondary diagnostic position. From 1993 through 2006, HZ incidence also increased 98 % from 1.7 (95%CI: 1.6-1.7) in 1993 to 4.4 (95 % CI: 4.3-4.4) in 2006. The increases occurred among all age strata and both sexes, although it increased more rapidly among females. Their results suggested greater increase in HZ rates between 1993 and 1996, prior to introduction of the varicella vaccination program. They also did not find any variation in HZ incidence by state varicella vaccination coverage rates. An interesting observation by the authors was that adults with dependents less than 12 years of age had lower HZ incidence at the outset of the varicella vaccination program compared with adults without dependents. However, the incidence in both groups became similar as the program progressed, suggesting that the introduction of the varicella vaccination program has not influenced HZ incidence in the general population, but it may have affected specific groups or households. In a recent Canadian epidemiologic study, Russell et al. [40] used multiple linked datasets to determine the incidence of HZ in Alberta, Canada and the impact of the varicella vaccination program, which has been publicly funded since 2002. Similar to our study, they showed that crude rates of medically attended HZ episodes increased over the interval of 1994–2010. Herpes zoster rate was 3.5 per 1000 person-years in 1994, 3.8/1000 person-years in 1998, 4.0/1000 person-years in 2001 and 4.5/1000 person-years by 2010. Rates were higher among females than males over the entire interval, and increased more rapidly for females than males. As in our study, prior to the publicly-funded varicella vaccination program in 2002, all age groups experienced increasing annual rates of HZ, but a sharp decline was seen in those less than 10 years of age for 2002–2010, the period in which varicella vaccination was publicly funded by the government. Other recent studies conducted in Australia [46], Japan [47] and Taiwan [48, 49] also have found that zoster rates were increasing before the introduction of their varicella vaccination programs. Although our study findings is in agreement with these studies, our data should be interpreted with caution. The varicella impact model we used may not capture the influence on zoster of varicella vaccination, because time periods we used were based on time of varicella vaccination policy changes and we did not allow for the impact of that vaccination on chickenpox rates to be seen. Because we did not allow for a lag time to see reductions in childhood varicella rates after introduction of a new vaccination program, we may have missed the true impact of childhood varicella vaccination program on HZ rates.

In our study we saw a very low PHN incidence of 0.343 per 1000 persons compared to the SPS trial [29], which showed the PHN incidence to be 1.38 per 1000 in the placebo group. In the systematic review conducted by Kawai et al., the risk of developing PHN ranged widely from 5 % to 30 % because of the different definitions used by researchers to classify duration, severity of pain and also because of differing comorbid illnesses, age and other underlying risk factors from one study to another [13]. The low incidence of PHN seen in our study is comparable to other studies using administrative databases to estimate PHN incidence; these studies are more likely to report a lower estimated risk of PHN compared with prospective studies (2.6 % to 6.9 %) [50‐53]. This is likely the result of misclassification bias when using billing codes rather than actual medical records to delineate PHN. Further, the definition of PHN varies widely from one study to another and therefore, the incidence varies depending on whether the study uses the 90 day or 30 definition, making the direct comparison difficult across studies. Regardless of the definition, we did see a significant increase in the incidence of PHN between 1997 and 2012. Given that the definition of PHN included a diagnosis of zoster, we had expected to see a rise in PHN incidence but the magnitude of a 3-fold increase was a surprise. Although some of this increase is likely real and the result of the increase in HZ rates, we feel it may also be related to increasing recognition and awareness by the public and medical community about PHN being a common complication of zoster. Like previous studies, we also showed that the highest incidence of PHN was in the older individuals, especially those over the ages of 70 years, but in terms of increasing incidence with time, we saw the largest increase in those 40–49 years and 10–19 years. Understanding of the risk factors for PHN has evolved over the years; not only are older age, greater pain and rash severity, and presence of a prodrome known to be risk factors for PHN, but newer evidence would suggest immunosuppression, diabetes and trauma to also be risk factors for PHN [4, 54]. It may be that the younger cohort are more immunosuppressed due to specific diseases or medications, or the proportion of young adults with diabetes is increasing, or perhaps they or their parents are more prone to seeking medical attention for PHN symptoms than the elderly.

Between 1997 and 2012, only 0.6 % (6 per 1000 population) of our HZ cases were hospitalized, and over time we saw a slight decrease in the hospitalization rate from 0.9 % in 1997 to 0.5 % in 2012. Studies that have reported on hospitalization rates for HZ show a wide variation with reported rates ranging from 2 to 25/100 000 person-years in studies examining all ages [13]; the variation is probably related to differing admission criteria and whether studies used the HZ primary code or all codes for diagnosis. In our study, we saw that the mean age of patients hospitalized was higher and in the mid-seventies compared to those being seen by their GPs. Although we did not breakdown our data by age, studies have shown that hospitalization rates increase with age. For example, Jackson et al. reported hospitalization rates of 10 per 100,000 in adults 60–69 years of age but 65 per 100,000 in adults 80 years old or above [55]. Rates of hospitalization, in an Australian study, showed similar increases from 13 per 100,000 to 96 per 100,000 in adults 60–64 years of age and those ≥80 years of age respectively [56].

In our dataset, despite seeing a slight decrease in the number of hospitalizations for the treatment of HZ, we saw a significant increase in hospitalization costs over time. We believe, this may be related to more complicated cases being admitted for treatment while uncomplicated cases are being treated as outpatients, as we also saw an increase in the number of cases seen per year by GPs with HZ. Ninety-five percent of our HZ patients were treated with antivirals, which is similar to rates seen in studies conducted in Germany (71 %) [57], Italy (79 %) [51], and France (94 %) [58]. In general per case treatment costs of inpatients with HZ are approximately $16,000 and outpatients is around $40 per GP visit (an additional $123 if you include medications). With respect to PHN, we saw that most PHN cases were being treated on an outpatient basis, again with an increase over time with respect to the number of cases being seen by GPs. Like HZ, we saw an increase in the treatment costs associated with PHN. Costs associated for PHN treatment in the hospital is approximately $14,000 per case and for outpatients is $53 per visit. Given the numbers of patients who have a diagnosis of PHN, the annual cost of visiting a GP and receiving treatment is $75,560 to the healthcare system. It is difficult to compare cost data to other studies as the approaches used to calculate costs are so different; some studies look at direct and indirect costs while others, like ours look at direct costs only; some look at outpatient visits only, while other studies look at all medical expenses. Gauthier et al. estimated the mean total cost in their study of inpatient to be £103 ($206) per HZ case and £397 ($794) per HZ case for outpatient care [36]. This is in contrast to our study which showed higher hospitalization costs rather than outpatient costs. In an Italian study, the mean inpatient treatment costs for HZ was €2592 ($4147) while outpatient costs were €123 ($197) [51]. While the trends are similar to our study – higher costs for hospital visits – our costs were much higher for HZ hospitalization ($16,389) but similar for outpatient costs ($160).

Although our data with respect to the increase in HZ and PHN rates over time agree with previously reported studies, this may be an artifact given that ours is an epidemiological study using large datasets, and we were not able to directly access the accuracy of the administrative database claims through record reviews. As with all studies using administrative data, we used billing codes meant for physician billing as our basis to delineate zoster and PHN diagnosis. Because of this, changes in coding or coding errors could have led to an over- or under-estimation of the results. To try and overcome some of these issues and increase specificity, we used alternative definitions and found similar results to our main analysis. Our study was limited by the data available in the databases so if patients did not approach a medical facility for their zoster infection or PHN treatment, we would not have captured these events or their costs in our analyses, thereby underestimating their incidence; this may be particularly true around PHN-related costs as patients could have had alternative treatments for pain control. It is also possible that increasing HZ and PHN rates are simply due to improved coding and awareness. This may be especially true after the herpes zoster vaccine was marketed in 2008 as both the public and the medical community have been better educated on signs and symptoms of herpes zoster, the use of antivirals and analgesics for its treatment and complications arising from zoster infection, particularly PHN. Around the time the zoster vaccine was marketed in Canada, pharmacists received the authority to immunize and one of the vaccines they are heavily promoting to their clients is the herpes zoster vaccine, further enhancing awareness of the disease. However having said that, we know that between 2008 and 2012, the herpes zoster vaccine was marketed as a freezer-stable product and the uptake rate of the vaccine was less than 10 %. Although we adjusted for age and sex, we are not able to control for all the potential confounders which may have affected the HZ and PHN rates. In particular, the proportion of the population which has comorbidities such as diabetes mellitus (a known risk factor for HZ), and immunocompromising conditions such as solid organ transplants, cancer, leukemia/lymphoma, or being on medications that decrease T-cells (TNF-alpha inhibitors, DMARDS, corticosteroids) has increased over time, leading to increased numbers of patients being at risk for HZ/PHN [59]. Finally, as with all epidemiological studies, there may be additional risk factors for zoster and PHN that we are unaware of at the present time and these would not have been adjusted for in our study.

In conclusion, we found the incidence of zoster and PHN is increasing with time. Although there are well known risk factors, such as age, female gender and being immunocompromised, which are contributing to the increased incidence seen with time, there are obviously risk factors present which are more difficult to tease out, such as trauma, psychological stress, race, and family history. Understanding the epidemiology and risk factors of HZ is critical for better targeting of treatment and prevention strategies.