JSTOR

Data Sources

The NIS is a component of the Healthcare Cost and Utilization Project, a set of healthcare databases sponsored by the Agency for Healthcare Research and Quality and developed by a partnership among federal and state governments and healthcare institutions. The NIS is the largest publicly available all‐payer inpatient database in the United States.

The NIS contains deidentified, patient‐level clinical and resource use data included in a typical discharge abstract. For each year, these data reflect hospital stays in approximately a 20% stratified sample of nonfederal community hospitals, including public hospitals, children’s hospitals, and academic medical centers but excluding long‐term hospitals, psychiatric hospitals, and chemical dependency treatment facilities. For the purposes of sampling the NIS, these US hospitals are divided into strata based on 5 hospital characteristics: US region, urban or rural location, ownership or control, bed size, and teaching status.

The study period was from 1993 through 2004. By starting with the year 1993, we were able to analyze 3 years of data before the first full year of national efforts at varicella vaccination in 1996. In 1993, data from 17 states were available; by 2004 (the most recent year for which NIS data were available), 37 states participated in the Healthcare Cost and Utilization Project. For each year of NIS data during the study period, discharge‐level sampling weights provided by the Agency for Healthcare Research and Quality were utilized by the investigators to derive nationally representative estimates of hospitalization rates standardized to the concurrent national population.¹³

US Census data for each of the years in the study period were used to obtain rate denominators for the overall population and for specific age groups of interest in the herpes zoster analyses and for varicella analyses.¹⁴ To interpret the analyses of discharge‐related hospital charges in constant dollars, we standardized all hospital charges to 2007 US dollars by use of the medical care component of the Consumer Price Index for each year of data.¹⁵

Identification of Hospital Discharges

Consistently with methods described in previous studies of cases of herpes zoster^9,11 and of herpes zoster–related hospitalizations,^16,17 we searched for diagnoses of herpes zoster (using the International Classification of Diseases, Ninth Revison, Clinical Modification [ICD‐9‐CM] codes starting with 053) in all 15 diagnostic‐code fields included for hospital discharges in the NIS during 1993–2004. We excluded from analyses hospitalizations with diagnostic codes for both herpes zoster and varicella (ie, ICD‐9‐CM codes starting with 052). All hospital discharges that were not categorized as HZHDs by use of these criteria were categorized as non‐HZHDs.

Clinical Classification Software, a tool developed by the Agency for Healthcare Research and Quality for grouping patient diagnoses and procedures into a manageable number of clinically meaningful categories, was used to determine the top 10 most frequent primary diagnoses among HZHDs for the years 1994, 1999, and 2004, to provide a measure of the balance of primary diagnostic categories during the entire study period.¹³ VRHDs were identified by use of similar diagnosis‐based methods, which were described in our previous study.⁶

Data Analysis

We designed our analysis to examine the rates of severe illness due to herpes zoster that resulted in hospitalization, as measured by the rates of HZHDs. The hypothesis was that, at a population level, the rates of HZHDs would increase over time as the rates of VZHDs decreased. The annual population‐adjusted rate of HZHDs (per 10,000 US population) and the annual inflation‐adjusted total charges for HZHDs were the primary outcomes. Secondary outcomes included mean charges for HZHDs and the distribution of total charges for HZHDs according to the expected primary payer. For the overall study period, we stratified data by age group, as we had for our earlier analysis of VZHDs.⁶ For the year 2004, we also included a specific age‐stratified analysis, to examine rates of HZHDs and associated charges for patients in the herpes zoster vaccine–eligible group aged 60 years or older.

We obtained descriptive statistics for hospitalizations and hospital charges as well as 95% confidence intervals (CIs) of the estimates. All results, unless specified otherwise, took into account the complex stratified sampling by use of sampling weights. SAS statistical software, version 8.2 (SAS) was used for all analyses.

Our study was funded internally, and all analyses were conducted by the authors. The authors had no financial interest in the outcome. Also, our study was exempt from review by the human subjects committee because of its analysis of deidentified secondary data.

NIS Characteristics

The NIS for the years 1993–2004 included data on 6.4–8.0 million unweighted hospital discharges annually, representing 34.6–38.6 million discharges nationally each year. The number of unweighted annual observations during the study period varied from 10,314 to 15,154 for HZHDs and from 661 to 3,270 for VRHDs.

Rates of Herpes Zoster–Related and Varicella‐Related Hospitalizations

Population‐adjusted rates of HZHDs did not change significantly from the prevaccination period (1993–1995) through the initial 5 years of the varicella vaccination period (Figure 1). Beginning in 2001, however, the rate of HZHDs overall began to increase, and by 2004 the overall rate was 2.5 HZHDs (95% CI, 2.38–2.62) per 10,000 US population, significantly higher than any of the rates calculated for the years prior to 2002.

(35 KB)

Figure 1.  Estimated annual rates of herpes zoster–related and varicella‐related hospital discharges (HZHDs and VRHDs, respectively) per 10,000 US population. The weighted point estimates for the rates of hospital discharge each year are shown in the trend line with corresponding 95% confidence intervals (whiskers) for herpes zoster and varicella.

Open New Window

Concurrently, population‐adjusted rates of VRHDs decreased significantly from 1993 through 2004 (Figure 1). The rate exceeded 0.5 VRHDs per 10,000 US population from 1993 to 1995 and declined significantly to 0.11 VRHDs (95% CI, 0.10–0.13) per 10,000 US population by 2004.

Age‐Related Patterns for HZHDs

The trend in the rate of HZHDs varied across age groups (Figure 2). Generally flat trends from 1993 through 1999 were subsequently followed by diverging patterns. HZHD rates among the elderly were significantly higher during 2002–2004 than during the years prior to 2001. HZHD rates for other age groups changed minimally by comparison.

(23 KB)

Figure 2.  Estimated annual population‐adjusted rates of herpes zoster–related hospital discharges (HZHDs) for specific age groups. The weighted point estimates for the rates of hospital discharge each year are shown, standardized to the year‐specific population for each age group. Corresponding 95% confidence intervals (whiskers) are shown for the group of patients aged 65 years or older.

Open New Window

Hospital Charges

Mean hospital charges for both HZHDs and non‐HZHDs increased significantly from 1993 to 2003 and then decreased slightly in 2004. For HZHDs, mean hospital charges were $22,790 (95% CI, $21,140–$24,441) in 1993 and $26,825 (95% CI, $25,266–$28,383) in 2004 (all in 2007 US dollars). Among non‐HZHDs, mean hospital charges were $17,204 (95% CI, $16,512–$17,894) in 1993 and $23,154 (95% CI, $22,228‐$24,080) in 2004.

Total annual hospital charges for HZHDs fluctuated between 1993 and 1999 and subsequently increased significantly (Figure 3). Overall, total annual hospital charges for HZHDs increased from $1.2 billion in 1993 to more than $1.9 billion by 2004 (all in 2007 US dollars). Among expected primary payers, Medicare accounted for the largest portion of annual total HZHDs charges, increasing by more than $0.5 billion, from $742 million in 1993 (61.3% of the total) to approximately $1.27 billion (ie, 66.2%) by 2004. During the same time period, Medicaid charges increased slightly from more than $143 million to more than $166 million, private‐payer charges increased significantly from approximately $249 million to more than $384 million, and charges from “other” payers increased from approximately $74 million to approximately $97 million. In contrast, total annual hospital charges for VRHDs declined significantly, from more than $230 million in 1993 to less than $81 million by 2004 (Figure 3).

(28 KB)

Figure 3.  Estimated weighted inflaction‐adjusted hospital charges for herpes zoster–related hospital discharges (HZHDs) and varicella‐related hospital charges (VRHDs). Whiskers, 95% confidence intervals.

Open New Window

Hospitalizations Among the Herpes Zoster Vaccine–Eligible Population

The herpes zoster vaccine–eligible population (ie, persons aged 60 years or older) accounted for 75% of HZHDs in 2004. The population‐adjusted rate was 11.23 HZHDs (95% CI, 11.03–11.45) per 10,000 US population, which was almost 6 times the next highest rate, among adults 50–59 years of age, which was 2.07 HZHDs (95% CI, 2.02–2.31) per 10,000 US population. The mean hospital charge for patients aged 60 years or older was $26,477 (95% CI, $24,957–$27,999), which did not differ substantively from the mean hospital charge for the entire study population. Total annual hospital charges for patients 60 years or older exceeded $1.4 billion and accounted for the majority (ie, 74%) of the total charges for patients of all ages in 2004.

Principal Diagnoses for HZHDs

In 1994, 1999, and 2004, the most common principal diagnosis category for HZHDs was viral infection, comprising 20.3%–25.7% of all HZHDs (Table). Pneumonia was consistently the second most common category (5.3%–6.3% of all HZHDs). The remaining 8 most common principal diagnoses varied slightly in frequency but showed no clear trend over the course of the study period (Table).

Table.  Rankings of the 10 Most Common Principal Diagnoses for Herpes Zoster–Related Hospital Discharges in the United States, for 3 Different Years of the 9‐Year Study Period

Open New Window

Age‐Specific Trends in the Varicella Vaccination Era

Brisson et al.⁹ used an age‐structured model of the natural history of varicella and herpes zoster to estimate the impact of varicella vaccination on the incidence of herpes zoster. During a 20‐year time horizon, they predicted that persons aged 10–44 years at the beginning of the time horizon would be most adversely affected as a result of varicella vaccination, because most of them (ie, those living with children) would not have their VZV‐specific cell‐mediated immunity boosted by their vaccinated children. Our analysis of the first 9 years of the US varicella vaccination program suggested a somewhat different pattern for herpes zoster–related hospitalizations, in which the most significant change was a large increase in the rate of hospitalizaton for patients aged 65 years or older. These findings contrast with the projections of Brisson et al.⁹ and suggest that reduced exposure to VZV may be as important at the community as well as family level, particularly for hospitalization patterns.

Incidences of herpes zoster and of herpes zoster–related hospitalization are comparatively low among children.^10,23 Edmunds and Brisson⁸ predicted that the incidence of herpes zoster in vaccinated cohorts of children should be lower than it was in unvaccinated cohorts (ie, in children cohorted before the varicella vaccination recommendation in 1995), because the varicella vaccine virus has been found to be less likely to reactivate than the wild‐type virus.²⁴ Although the trends were modest in magnitude, we found that HZHD rates trended down toward the end of the study period for children in 2 age groups: 0–4 and 5–9 years. These findings suggest a benefit for young children from varicella vaccination, in terms of reductions in both varicella‐related and herpes zoster–related hospitalizations.

Although our findings are limited to patient hospitalizations and, therefore, are influenced by generally higher hospitalization rates for older adults than for younger adults, our study corroborates other analyses of herpes zoster that consistently document a sharp increase in the likelihood of contracting herpes zoster with increasing age.^3,11,17 Jumaan et al.¹⁰ also found that older populations had higher incidence rates of herpes zoster than did other age groups.

Economic Implications of Varicella Vaccination

In our study, we found that reductions in VRHDs were associated with annual savings of approximately $150 million in hospital charges, compared with the pre–varicella vaccination era. This finding extends our previous analysis, in which we found annual savings of more than $100 million in annual hospital charges related to primary varicella by 2001.⁶

However, our results also indicate that savings from reduced VRHD rates may be potentially overshadowed by a concurrent $700 million increase in annual hospital charges for HZHDs. In particular, we found that the herpes zoster vaccine–eligible population (persons aged 60 years or older) accounted for 74% of the total annual hospital charges in 2004, including Medicare charges of more than $1.26 billion annually for HZHDs, an increase of more than $500 million annually, compared with 1993.

Implications for VZV Vaccination Efforts

Varicella vaccination has had a tremendously favorable impact on reducing the incidence of primary varicella, but our results have shown that there has been a concurrent increase in HZHDs. Brisson et al.⁹ estimated that the overall incidence of herpes zoster will increase beginning in the first decade of varicella vaccination and then decline once the vaccinated cohort reaches middle adulthood, approximately 46 years after varicella vaccination began. In other words, although varicella vaccination may have negative consequences in terms of an increasing incidence of herpes zoster in the short run, it is expected to be beneficial in the long run. Nine years into the US childhood varicella vaccination program, it is premature to judge whether such a pattern will emerge.

The vast majority of cases of herpes zoster occur in older adults, and therefore the herpes zoster vaccine newly recommended for adults¹² may strongly affect future patterns of herpes zoster–related hospitalizations and associated charges. One case of herpes zoster can be prevented by vaccinating just 17 elderly adults,²⁵ but several barriers may deter patients from acquiring vaccination. Chief among the potential deterrents may be the cost of the herpes zoster vaccine. The vaccine costs approximately $150 per dose, which is almost double the next highest‐priced adult vaccine.²⁶ The comparatively high cost is complicated by the fact that coverage for the herpes zoster vaccine depends on prescription drug plans under Medicare Part D, in contrast to the pneumococcal and influenza vaccines, which are covered under Medicare Part B.

Published models of the cost‐effectiveness of the herpes zoster vaccine have concluded that cost‐effectiveness depends heavily on the cost of the vaccine and on the age of those vaccinated, and that the cost‐effectiveness of the herpes zoster vaccine exceeds $100,000 per quality‐adjusted life‐year under most conditions.^27,28 The charges for herpes zoster–related hospitalization used in the published base‐case estimates (when converted to hospital charges using a conventional cost‐to‐charge ratio of 0.5 and inflated to 2007 US dollars) ranged from approximately $14,350 to $15,600.^27,28 These values are less than the mean hospital charge of $26,825 that we found for HZHDs, which means that the authors of the cost‐effectiveness analyses may have underestimated the savings from the herpes zoster vaccination attributable to decreased herpes zoster–related hospitalizations. Therefore, to the extent that cost‐effectiveness estimates are influenced by hospitalization charges, the herpes zoster vaccine may be more cost‐effective than previously thought.

Study Limitations

As with all analyses of hospital discharge data, our findings must be interpreted with the caveat that there may have been the inadvertent exclusion or misattribution of ICD‐9‐CM codes pertinent to our study. In particular, it is possible that primary varicella and herpes zoster diagnoses may have been confused, either at the bedside or in hospitals’ coding efforts. For this reason, on the basis of epidemiologic patterns, some investigators have reclassified apparent VRHD in persons aged 50 years or older as HZHD.^5,21 We assessed our data to determine whether such reclassification would have materially changed our conclusions, and we found that it did not.

As with all cross‐sectional rather than longitudinal data, we cannot definitively draw a causal link between national varicella vaccination efforts and time trends in hospitalizations and charges for varicella‐related and herpes zoster–related disease. Rather, we have described a strongly suggestive association specific to the time periods before and after the initiation of the national childhood varicella vaccination program.

Our analysis centers on the incidence of severe illness due to herpes zoster infection that resulted in hospitalization, and we cannot conclude with certainty that a similar trend is present for incidence of herpes zoster more generally. In addition, hospital charges could not be disaggregated to characterize the economic effects of herpes zoster in the hospital setting. Rather, we present total charges for all hospitalizations, with herpes zoster as a contributing diagnosis, to reflect the economic toll of hospitalizations in which herpes zoster was present in a clinically significant manner.

Conclusions

Childhood varicella vaccination has been associated with profound reductions in the rate of VZV disease and in associated healthcare expenditures in the United States, but the long‐term effects of childhood varicella vaccination on the incidence of herpes zoster will not be known for several years. Concerns that increasing immunity against VZV, at the community level, may result in an increase in the incidence of herpes zoster are justified on the basis of the first 9 years of national hospitalization data following the initiation of national varicella vaccination efforts.

The majority of herpes zoster–related hospitalizations and hospital charges are attributable to adults in the herpes zoster vaccine–eligible group (persons aged 60 years and older). Therefore, a herpes zoster vaccine has the potential to mitigate the recent increase in the rate of hospitalizations due to herpes zoster. Potential barriers to herpes zoster vaccination must be addressed, to promote nationwide immunization efforts among older adults.