JAMDA 13 (2012) 54e59
JAMDA journal homepage: www.jamda.com
Economic Burden of Herpes Zoster Among Skilled Nursing Facility Residents in the United States Liyuan Ma PhD, MA a, *, Ronald R. White MS, MA a, Siva Narayanan MS, MHS b, Kenneth E. Schmader MD c a
Merck & Co., Inc., Whitehouse Station, NJ TNS Healthcare, New York, NY c Center for the Study of Aging and Human Development, Division of Geriatrics, Department of Medicine, Duke University Medical Center; and Geriatric Research, Education and Clinical Center (GRECC), Durham Veterans Affairs Medical Center, Durham, NC b
a b s t r a c t Keywords: Herpes zoster skilled nursing facilities aged cohort study economic burden
Objectives: To evaluate health care resource use and direct medical costs attributable to herpes zoster (HZ) among elderly residents of skilled nursing facilities (SNF). Methods: This was a retrospective matched cohort study using data from 300 SNF in the United States. A total of 404 patients with HZ were matched with 1616 patients with no documented HZ or post-herpetic neuralgia using propensity scores. The study period included a 1-month pre-index period and the 90-day acute/subacute phase following the HZ index date. Health care resource utilization was captured from the long term care minimum dataset (MDS) and SNF admission-discharge records. The direct medical costs consisted of the SNF net bed revenue and hospitalization cost. Results: Over the 4-month study period, signiﬁcantly more patients were hospitalized among the HZ (20.5%) than non-HZ cohort (14.4%). Both the numbers of hospitalization episodes and hospitalization days were greater for HZ than for non-HZ patients. An average additional 0.09 hospitalization episodes and 0.55 days of hospitalization were estimated in a multivariate model for patients in the HZ compared with the non-HZ cohort. The incremental direct medical costs, which are composed of the incremental medical costs incurred in the SNF, and the incremental costs attributable to hospitalization, were estimated between $1079 and $1673 for patients with HZ. Conclusion: In the SNF setting, the presence of HZ imposes signiﬁcant health care resource utilization and direct medical costs. Copyright Ó 2012 - American Medical Directors Association, Inc.
Herpes zoster (HZ), caused by the reactivation of varicellazoster virus (VZV) from a latent infection in the spinal dorsal root or cranial sensory ganglia, is usually manifested by a unilateral vesicular rash and pain.1,2 The main complications of HZ are acute and chronic neuropathic pain or postherpetic neuralgia (PHN),3,4 which can last for months, even years.3,4 Approximately 1 million Americans develop HZ each year.5 The incidence and severity of HZ, as well as the frequency and severity of its complications, increase with age.5,6 In the United States, the incidence of HZ in individuals younger than 40 years has been estimated to range from 1.1 to 1.9 cases per 1000 patient-years, after This study was made possible by a grant from Merck. This article was prepared with the assistance of BioMedCom Consultants inc., Montreal, Canada. K.E.S. received Merck research grants for this study and other research activities. S.N. received Merck research grants to conduct this study and other research activities. L.M. and R.W. are employees of Merck, the study sponsor, and may hold stock options. * Address correspondence to Liyuan Ma, PhD, MA, Global Outcomes Research, Merck & Co., Inc., PO Box 4, WP97A-243, West Point, PA 19486. E-mail address: [email protected]
which it begins to rise, with 2.3 to 2.9 cases per 1000 patient-years among individuals aged 40 to 49 years and up to 12.0 cases per 1000 patient-years among individuals aged 80 years or older.4,7 Similar to that for HZ, the incidence and prevalence of PHN increase greatly with age.8,9 In a US study by Choo and colleagues,10 the prevalence of PHN was 14.7-fold higher (95% conﬁdence interval [CI]: 6.8e32.0) at 30 days after HZ onset and 27.4-fold higher (95% CI: 8.8e85.4) at 60 days among individuals aged 50 years or older compared with younger patients. The results of a recent US study by Yawn and colleagues,4 where PHN was deﬁned as zoster-associated pain persisting for at least 90 days, shows that PHN occurred in 18% of 1669 adult patients with HZ and in 33% of 246 patients aged 79 years and older. Overall, 10% of all patients with HZ experienced one or more nonpain complications such as ocular and neurological complications or skin superinfections.4 The substantial burden of illness that HZ and PHN produce in community-dwelling outpatients is well documented; HZ, and more particularly PHN, have been shown to interfere with activities of daily living, to lower health-related quality of life, and to worsen
1525-8610/$ - see front matter Copyright Ó 2012 - American Medical Directors Association, Inc. doi:10.1016/j.jamda.2010.03.015
L. Ma et al. / JAMDA 13 (2012) 54e59
physical and mental health.11,12 To reﬂect personal burden associated with HZ, a recent study focusing on the value attributed by the community and patients to HZ prevention reported that community members “would trade substantial amounts of time or money to avoid herpes zoster, even in the least severe scenarios.”13 Researchers have attempted to evaluate health care use and the economic burden related to HZ in various US settings. Coplan and colleagues14 indicated that, in 1994 among Kaiser Permanente Medical Care Program (Northern California, US) members, HZ resulted in 2.1 hospitalizations per 100,000 patient-years; this number encompassed hospitalization for HZ across all age groups. However, hospitalization was far more common among adults aged 60 years or older than in those aged 25 through 59 years. More recently, 2 US studies showed signiﬁcant incremental health care use during the acute/subacute HZ period (deﬁned as the 21 days preceding and 90 days following initial HZ diagnosis) among patients in a managed care setting. Cost estimates in these studies ranged from $431 to $605, with most of the cost burden caused by outpatient care and prescription drug costs.15,16 Interestingly, the average estimated incremental costs per patient with acute/ subacute disease among individuals aged 80 years or older were more than 3 times that of individuals aged 19 years or younger.15 Both the aforementioned study by White and colleagues16 and an additional US study17 estimated HZ-related medical costs in the ﬁrst year after HZ diagnosis compared with those for a control nonHZ population, which were propensity score matched for demographics and comorbidities. The average incremental medical costs were estimated to range from $757 to $1313 for HZ cases and from $2292 to $5742 for PHN cases.16,17 There is empirical evidence highlighting the substantial clinical and economic burden of HZ in the US population; however, much of this research has concentrated on managed care and outpatient settings. Although the burden of HZ has proven to be high among older people, there are sparse data on burden of HZ among longterm care (LTC) cohorts.18 Further, because the incidence of HZ in older people from community-dwelling populations is itself often estimated, the incidence of HZ in the LTC subpopulation of older people is not known with certainty. With approximately 1.5 million elderly residents in US nursing homes, the occurrence of at least 10,500 to 16,500 HZ cases may be estimated per year using community-based studies, which report an incidence of 7 to 11 cases of HZ per 1000 elderly persons per year.18 It has also been suggested that residents of LTC facilities are at greater risk for developing HZ because of frailty; therefore, the number of HZ cases among residents of nursing homes may be underestimated.18 The objective of this study was to evaluate health care resource utilization (HCRU) and the direct medical costs of HZ for patients cared for in skilled nursing facilities (SNF). Methods Study Design and Patients This was a retrospective matched cohort study of herpes zoster (HZ cohort) patients and non-HZ patients (control cohort) who
received care in 1 of 300 skilled nursing facilities of a large LTC provider from 2002 to 2007. The centralized clinical and administrative data were available for these patients and the clinical and resource use data were extracted from the minimum data set (MDS).19 The MDS is part of the federally mandated process for clinical assessment of all residents in Medicare- or Medicaidcertiﬁed nursing homes. This process provides a comprehensive assessment of each resident’s functional capabilities and helps nursing home staff identify health problems. Resident Assessment Protocols (RAPs) are part of this process, and provide the foundation on which a resident’s individual care plan is formulated. MDS assessment forms are completed for all residents in certiﬁed nursing homes, regardless of source of payment for the individual resident. MDS assessments are required for residents on admission to the nursing facility and then periodically, within speciﬁc guidelines and time frames. HZ cases were ﬁrst identiﬁed using the diagnosis (ICD-9) database and admission/discharge information. Subjects were included if they had a clinical diagnosis of HZ (ICD-9 diagnosis [primary or nonprimary] ¼ 053.xx), had at least 90 days of follow-up data starting with the date of ﬁrst HZ diagnosis, and had at least 180 days of baseline data before the ﬁrst HZ diagnosis. This 180-day period without an HZ diagnosis was used to ensure that incident cases were analyzed and that existing cases were excluded. Patients who had no documented diagnosis of HZ and PHN were considered controls, and were required to have at least 9 months of continuous enrollment. The date of HZ diagnosis was used as the index date for HZ cases. For control patients who had no documented diagnosis of HZ and PHN at anytime during their stay at an SNF, a random index date was assigned within the duration of the patient’s SNF stay, determined after excluding the initial 180 days from admission and 90 days before discharge. Data were captured for the prodromal/acute/ subacute phase of HZ, deﬁned as the 30 days preceding (prodromal phase) and 90 days following (acute/subacute phase) the index date, for a total of 120 days, hereafter referred to as the “observation period”. The baseline period encompasses the 150 days preceding the prodromal phase (see schematic). The original sample for matching was composed of 426 HZ patients and 26,365 non-HZ patients who met inclusion criteria. A logistic-regression model was constructed to assign to each patient a propensity score deﬁned as the predicted probability that this patient belongs to the HZ group. Predictor variables including demographics (age, gender, ethnicity), Charlson comorbidity index, total number of comorbid conditions (from MDS), and other selected characteristics (presence of speciﬁc other comorbidities such as pressure ulcers, urinary tract infections, depression, and schizophrenia; pain symptoms; resident falls/fractures; past medication history such as use of anticonvulsants, barbiturates, antipsychotics; SNF length of stay; baseline nursing labor cost; and net bed revenue) were entered into the model. Each patient in the HZ group was matched with 4 individuals in the non-HZ group who had the most similar propensity score, with the requirement that any difference between the scores of the persons in a pair did not exceed a preset limit of 0.01.
L. Ma et al. / JAMDA 13 (2012) 54e59
The study protocol was reviewed and approved by IntegReview, an independent institutional review board located in Austin, Texas. Outcome Measures Hospitalization of SNF patients is captured in the discharge/ admission database that records each incidence of discharging patients to hospitals and their return to the SNF. Episodes not resulting in an overnight stay at the hospital were considered as emergency room visits and were not included under direct medical costs because of their low occurrence. Both the number of hospitalization episodes and hospitalization days were analyzed. Although recorded in the MDS, because physician visits apply only to the 14 days preceding the latest entry into the MDS, the precise number of these visits within the SNF setting for the study evaluation period could not be captured and was therefore not included in the ﬁnal economic evaluation. Incremental medical costs were estimated in 2 parts. The ﬁrst part estimates incremental costs incurred by patients with HZ while in the skilled nursing facility; the second part estimates the incremental cost of hospitalizations for patients with HZ. Incremental costs incurred by patients with HZ while in the skilled nursing facility were estimated from net bed revenue because reimbursements for the skilled nursing facilities were not based on the actual services provided and actual medical costs were not available. The net bed revenue, which is the monthly reimbursement received by an LTC provider for all services provided for patients based on their health status, was used as a surrogate measure for the medical costs and was adjusted to 2007 dollars using the medical consumer price index. Incremental hospitalization costs were estimated by multiplying the average costs per hospitalization (from the literature) and the difference in the number of hospitalizations between HZ and non-HZ patients. The average hospitalization costs per episode attributable to HZ for the older adult population ranged from $6815 to $13,412.20e23 Using this range, we calculated a lower and upper bond for the incremental cost of hospitalization attributable to HZ for our study population. The total direct medical costs consisted of both net bed revenue and costs attributable to hospitalization. Statistical Analyses Continuous variables were summarized using mean and standard deviation and sample means were compared using t tests between the HZ and control cohorts. Categorical variables were summarized using the sample size and proportions, with the comparison between cohorts based on the chi-square test. End points such as hospital days and costs had a skewed, long right-tail distribution. Hence, they were analyzed using a generalized linear model,24 which accommodates more ﬂexible meanvariance structures. With the emphasis of the study being the estimation of incremental difference between HZ and non-HZ cohorts, rather than the identiﬁcation of risk factors, the ﬁnal multivariate model included all clinically important baseline prognostic factors, such as age, gender, race, activities of daily living scores, pain severity at baseline, baseline costs, and an indicator variable for being hospitalized in the baseline period. To account for the possible overdispersion and its impact on the statistical inference, a negative binomial distribution with log link was used in lieu of the Poisson distribution.25 The negative binomial distribution is a generalization of Poisson regression, which incorporates overdispersion, allowing the variance to be a quadratic function of the mean. The model’s goodness of ﬁt was assessed via the deviance chi-square statistic. Sensitivity analyses
were performed using a linear model to assess the robustness of the negative binomial results. In the generalized linear model, the adjusted means for the HZ and control non-HZ patients were based on actual distribution of the sample population rather than the default equal weight distribution. Because a substantial number of patients were not being hospitalized, the hospitalization days were analyzed via a 2part model: the ﬁrst part was to estimate the probability of having a length of stay greater than zero for patients in the HZ and non-HZ cohorts, and the second part was to estimate the length of stay using a multivariate generalized linear model for the HZ and nonHZ patients who were hospitalized. Results Baseline Characteristics Table 1 shows the characteristics at baseline from a total of 2020 patients consisting of 404 patients who later developed HZ and 1616 control non-HZ patients matched according to propensity scores. HZ patients ranged in age from 27 to 100þ years, with a mean age of 82.6 years and matched non-HZ control patients from 41 to 100þ with a mean age of 82.9 years (Table 1). Nearly all patients (91.3% and 89.9% for HZ and non-HZ patients, respectively) were 70 years or older and about three quarters (76.2% and 75.7% for HZ and non-HZ patients, respectively) were female. White patients accounted for 90.8% and 92.8% of the HZ and matched nonHZ cohort, respectively; 0.2% to 6.2% of HZ patients and 0.2% to 4.6% of non-HZ control patients were from other ethnicities. During the 150-day baseline period, 71% of the HZ patients and 69.8% of the matched controls did not report pain. The mean Charlson comorbidity index at baseline was 0.31 for the HZ population and 0.30 for the matching non-HZ population. The mean baseline score for activities of daily living was 19.3 for either cohort. At baseline, 21.3% of patients in the HZ cohort and 18.8% of patients in the non-HZ cohort had been hospitalized; the mean numbers of hospitalization days for those hospitalized were 7.95 and 8.79, respectively. There were no statistically signiﬁcant differences in either the modes of locomotion or transfer, or in number of falls during the preceding 30 or 180 days between patients in the HZ and non-HZ cohorts. Patients in the matched HZ and non-HZ cohorts were similar for all demographics and baseline characteristics such as number of emergency room episodes, physician visits, or net bed revenue. HCRU and Net Bed Revenue Associated with HZ The proportion of patients who were hospitalized during the 4month observation period in the HZ cohort (20.5%) was signiﬁcantly higher than in the non-HZ cohort (14.4%; P ¼ .002) (Table 2). In addition, the number of hospitalization episodes was signiﬁcantly greater among HZ patients than non-HZ patients (0.28 vs. 0.19, P¼.002). Among the cohort of hospitalized patients (N¼316), 72% of HZ patients and 73% of non-HZ patients had only one episode of hospitalization. The mean length of hospitalization stay for the hospitalized cohort was 8.35 days for the HZ patients and 8.74 days for non-HZ patients. For the overall study population (N ¼ 2020), the mean length of hospitalization stay was 1.72 days and 1.26 days for the HZ and control cohorts, respectively. The number of physician visits, captured for 14 days before the last MDS update, was signiﬁcantly higher among HZ than among non-HZ patients. The average net bed revenue, which does not cover the costs incurred while the patient is hospitalized, was $19,074.50 for the HZ patients and $18,474.80 for the non-HZ patients during the observation period.
L. Ma et al. / JAMDA 13 (2012) 54e59
Table 1 Demographics and Baseline Characteristics (During the 150-Day Baseline Period) Parameter
HZ Cohort (n ¼ 404)*
Non-HZ Cohort (n ¼ 1616)*
Age at index (mean SD) Gender Male (%) Female (%) Race White (%) Black (%) Hispanic (%) Asian/Paciﬁc (%) American Indian/Alaskan (%) Pain symptoms Pain free (%) Less than daily/mild (%) Less than daily/moderate (%) Less than daily/excruciating (%) Daily and mild (%) Modes of locomotioneNo tools/aid used (%) Modes of transfereNone used (%) Falls in past 30 days (%) Falls in past 180 days (%) Hospitalized Yes (%) No (%) Number of hospitalization episodes (mean SD) Number of hospitalization days (mean SD) Number of hospitalization days for those hospitalized (mean SD) Number of ER episodes (mean SD) Number of physician visits (mean SD)x Activities of daily living self assessment score (mean SD) Charlson Comorbidity Index (mean SD) Baseline Net Bed Revenue (mean SD) Number of long term care years before index date (mean SD)
96 (23.8) 308 (76.2)
393 (24.3) 1223 (75.7)
367 25 7 4 1
(90.8) (6.2) (1.7) (1.0) (0.2)
1499 75 22 17 3
(92.8) (4.6) (1.4) (1.1) (0.2)
287 43 52 2 20 153 19 47 115
(71.0) (10.6) (12.9) (0.5) (5.0) (37.9) (4.7) (11.6) (28.5)
1128 170 208 20 90 610 96 202 437
(69.8) (10.5) (12.9) (1.2) (5.6) (37.8) (5.9) (12.5) (27.0)
86 (21.3) 318 (78.7) 0.27 0.59 1.69 4.59 7.95 7.05 0.01 0.11 0.61 0.73 (N¼350) 19.30 10.16 0.31 0.30 $22,763.60 $6370.15 2.87 2.37
303 (18.8) 1313 (81.3) 0.27 0.65 1.64 5.57 8.79 10.19 0.01 0.09 0.59 0.74 (N¼1366) 19.30 10.00 0.30 0.29 $22,453.00 $6316.02 2.65 2.38
.960z .340z .640z .570z .247z .848y .851y .477y .330y .652y .876y .441y .378y .096y
Activities of daily living assessed: bed mobility, walk in the room, walk in the corridor, locomotion off unit, dressing, eating, toilet use, and personal hygiene. Pain symptoms are measured in MDS in terms of frequency and intensity. The pain measures correspond to the highest level of pain present in the past 7 days. Modes of locomotion include cane/walker/crutch, self-wheeling, wheeling by another person, wheeling as a primary mode of locomotion. Modes of transfer include bedfast all or most of the time, use of bed rails for bed mobility or transfer, manual lifting, mechanical lifting, transfer aid (eg, slide board, trapeze, cane walker), or no assistance needed. Incidence of falls is captured under the category of “accidents” in the MDS. *n ¼ number of patients in the cohort unless otherwise speciﬁed. y t test. z Chi-square test. x Occurring during the 14 days preceding minimum data set (MDS) form completion.
Results from the Multivariate Model The adjusted mean numbers of hospital episodes from the multivariate model were 0.25 and 0.16 for patients in the HZ and non-HZ cohorts, respectively, with a difference of 0.09 (Table 3). The mean incremental costs of hospitalization were estimated at $613 to $1207 for patients with HZ (incremental number of hospitalizations multiplied by the average cost of hospitalizations in older adults from the literature). The adjusted mean length of hospitalization stay was 1.70 days for HZ patients and 1.15 days for non-HZ patients with a difference of 0.55 days (ratio: 148%). The adjusted mean net bed revenue was $18,813 for the HZ patients and $18,347 for the non-HZ patients during the observation period, resulting in average incremental costs of $466 for patients with HZ; this represents only costs incurred in the SNF, but does not include costs incurred while a patient is hospitalized. Overall, we estimated that the direct medical costs for HZ patients were $1079 to $1673 higher than non-HZ patients in the 4-month observation period. Discussion During the 4-month observation period, there were signiﬁcantly more patients hospitalized among the HZ cohort than among the control non-HZ cohort. In a recent population-based study in the US, HZ-related hospitalizations were reported in 3% of HZ patients
across all ages and 9% of patients aged 80 years or older.4 The current study, evaluating a population requiring long term care and taking into consideration all hospitalizations whether speciﬁcally HZ related or not, reported a higher proportion, 20.5%, of HZ patients requiring hospitalization compared with 14.4% in the nonHZ patients. The average number of hospitalization episodes and days were also signiﬁcantly greater among the HZ than the non-HZ cohort. Interestingly, a sharp increase in the number of hospitalization episodes attributable to HZ, depending on patient age, was previously shown in the US, rising from approximately 14 hospitalizations per 100,000 patients aged 50 to 59 years to approximately 190 per 100,000 patients aged 80 years or older.26 Similarly, even though physician visits in the MDS nursing records were captured for only 14 days before the last update, the number of physician visits was signiﬁcantly higher among HZ than among non-HZ patients. In a recent UK study, the number of general physician visits for HZ patients aged 50 years and older, using a large primary-care dataset and ofﬁcial hospital episode statistics, reported a mean number of 1.4 visits per HZ episode.27 The incremental cost estimate of $1079 to $1673 for the observation period (1 month preceding through 3 months after the HZ index date) is considerably higher than previously published cost estimates in the general population. By comparison, 2 US studies estimated direct medical costs attributable to HZ at $431 and $605 during the 21 days preceding and 90 days following initial HZ
L. Ma et al. / JAMDA 13 (2012) 54e59
Table 2 Summary Statistics of Outcome End Points (During the 120-Day Observation Period) HZ Cohort (n ¼ 404)*
Parameters Hospitalized Yes (%) No (%) Number of hospitalized episodes (%) 0 1 2 3 4 5 Number of hospitalization episodes (mean SD) Number of hospitalization days (mean SD) Number of hospitalization days for those hospitalized (mean SD) Admitted to ER (%) Yes No Number of ER episodes (mean SD) Physician visit (%)x Missing Yes No Number of physician visits (mean SD)x Net bed revenue (mean SD)
Non-HZ Cohort (n ¼ 1616)*
83 (20.5) 321 (79.5)
233 (14.4) 321 (79.5)
321 (79.5) 60 (14.9) 17 (4.2) 4 (1.0) 1 (0.2) 1 (0.2) 0.28 0.65 1.72 5.10 8.35 8.47 (n ¼ 83)
1383 (85.6) 170 (10.5) 51 (3.2) 11 (0.7) 1 (0.1) 0 (0.0) 0.19 0.52 1.26 5.55 8.74 12.19 (n ¼ 233)
3 (0.7) 401 (99.3) 0.01 0.09
2 (0.1) 1614 (99.9) 0.00 0.04
5 177 (43.9) 226 (56.1) 0.60 0.80 (n ¼ 403) $19,074.5 $5491.12 (n ¼ 393)
1 623 (38.7) 988 (61.3) 0.48 0.70 (n ¼ 1611) $18,474.8 $4599.0 (n ¼ 1555)
P Value .002z
.002y .135y .786y .025z .025y .054z
¼ number of patients in the cohort unless otherwise speciﬁed. t test. z Chi-square test. x Occurring during the 14 days preceding minimum data set (MDS) form completion.
diagnosis, and during the ﬁrst 90 days following HZ diagnosis, respectively.15,16 Because both studies used insurance claims from all ages, the observed disparity compared with our results may be attributable to the difference in average age between the groups studied. The frail health status of SNF residents could also contribute to the higher costs of HZ in our study, beyond that which is attributable to more advanced age alone. The most elderly age group for which costs were given in the study by Insinga and coworkers15 was over 80 years old. This group incurred incremental costs attributable to HZ of $805. In the other study, costs were not stratiﬁed by age for the acute/subacute phase of disease.16 In the present study, we did not have sufﬁcient cases remaining at 1 year post-HZ index date to analyze costs for this time period. However, White and colleagues16 demonstrated that substantial HZ-related costs were incurred in the chronic phase (beyond 3 months post-HZ index) with total HZ-related costs of $1052 after 1 year (inclusive of the acute/subacute phase deﬁned as the ﬁrst 90 days following HZ diagnosis) compared with $605 for the acute/ subacute phase alone. Therefore, the HZ-related costs in the SNF are likely to be substantially more by including the chronic phase than the estimates of $1079 to $1673 for the 4-months observation period. There are a number of limitations to our study. HZ cases were identiﬁed via ICD-9 coding, which may not always be accurate.28 Some data on several items that may be expected to inﬂuence costs for HZ care such as physician evaluations, emergency room
visits, or pharmacy costs have not been accounted for. Although emergency room visits were captured, they were not included in the direct medical costs because of their low prevalence. Because the MDS question applies only to the 14 preceding days, the exact count of physician visits could not be evaluated over the follow-up period; the use of extrapolation was avoided. However, signiﬁcantly more physician visits were reported for the HZ cohort within a given 14-day window implying costs that are separately unaccounted for in this research. “Bed-hold,” a common practice within nursing home settings in anticipation of patient’s return from hospital corresponding to a loss of bed revenue for the nursing home system, was not considered. The noninclusion of these parameters may potentially contribute to underestimation of total direct medical costs. Net bed revenue has been used as a surrogate marker for burden of care in the SNF setting, because of the absence of data on reimbursement levels corresponding to the individual services rendered to an SNF resident during his or her stay. This indirect method of cost attribution to SNF care may have added some level of approximation to the assessment of economic burden. Last, the costs per HZ hospitalization are obtained from previously published studies,20e23 and are not measured directly in this study. However, these published studies focused on patients who do not require skilled nursing care and are probably less frail and medically vulnerable. Therefore, the HZ hospitalization costs estimates used in the current analysis might underestimate the true costs for the nursing home population.
Table 3 Multivariate Analysis Results on Hospitalization and Costs (During the 120-Day Observation Period) Parameter
Hospitalization episodes Hospitalization days Net bed revenuey
HZ Cohort vs Non-HZ Cohort
(404) (404) (393)
0.25 1.70 $18,813.00
(1616) (1616) (1555)
0.16 1.15 $18,437.00
0.09 0.55 $465.70
.001 NA* .032
NA, not applicable. *No P value was generated for the hospitalization days from the 2-part model to accommodate a substantial number of patients not being hospitalized. y Surrogate for costs within long term care in US$.
L. Ma et al. / JAMDA 13 (2012) 54e59
In conclusion, this study demonstrates that the presence of HZ is associated with a substantial economic burden in SNF settings. Future research is needed to establish optimal strategies for HZ treatment,29 especially with regard to pain management, and for HZ prevention via vaccination30 among older people in LTC environments.
References 18. 1. Dworkin RH, Johnson RW, Breuer J, et al. Recommendations for the management of herpes zoster. Clin Infect Dis 2007;44(Suppl 1):S1e26. 2. Gnann JW Jr, Whitley RJ. Clinical practice. Herpes zoster. N Engl J Med 2002; 347(5):340e346. 3. Johnson RW. Consequences and management of pain in herpes zoster. J Infect Dis 2002;186(Suppl 1):S83eS90. 4. Yawn BP, Saddier P, Wollan PC, et al. A population-based study of the incidence and complication rates of herpes zoster before zoster vaccine introduction. Mayo Clin Proc 2007;82(11):1341e1349. 5. Oxman MN, Levin MJ, Johnson GR, et al. A vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. N Engl J Med 2005;352(22):2271e2284. 6. Donahue JG, Choo PW, Manson JE, et al. The incidence of herpes zoster. Arch Intern Med 1995;155(15):1605e1609. 7. Insinga RP, Itzler RF, Pellissier JM, et al. The incidence of herpes zoster in a United States administrative database. J Gen Intern Med 2005;20(8): 748e753. 8. Edmunds WJ, Brisson M, Rose JD. The epidemiology of herpes zoster and potential cost-effectiveness of vaccination in England and Wales. Vaccine 2001; 19(23e24):3076e3090. 9. Hope-Simpson RE. Postherpetic neuralgia. J R Coll Gen Pract 1975;25(157): 571e575. 10. Choo PW, Galil K, Donahue JG, et al. Risk factors for postherpetic neuralgia. Arch Intern Med 1997;157(11):1217e1224. 11. Oster G, Harding G, Dukes E, et al. Pain, medication use, and health-related quality of life in older persons with postherpetic neuralgia: Results from a population-based survey. J Pain 2005;6(6):356e363. 12. Schmader KE, Sloane R, Pieper C, et al. The impact of acute herpes zoster pain and discomfort on functional status and quality of life in older adults. Clin J Pain 2007;23(6):490e496. 13. Lieu TA, Ortega-Sanchez I, Ray GT, et al. Community and patient values for preventing herpes zoster. Pharmacoeconomics 2008;26(3):235e249. 14. Coplan P, Black S, Rojas C, et al. Incidence and hospitalization rates of varicella and herpes zoster before varicella vaccine introduction: A baseline assessment
24. 25. 26.
28. 29. 30.
of the shifting epidemiology of varicella disease. Pediatr Infect Dis J 2001;20(7): 641e645. Insinga RP, Itzler RF, Pellissier JM. Acute/subacute herpes zoster: Healthcare resource utilisation and costs in a group of US health plans. Pharmacoeconomics 2007;25(2):155e169. White RR, Lenhart G, Singhal PK, et al. Incremental 1-year medical resource utilization and costs for patients with herpes zoster from a set of US health plans. Pharmacoeconomics 2009;27(9):781e792. Dworkin RH, White R, O’Connor AB, et al. Healthcare costs of acute and chronic pain associated with a diagnosis of herpes zoster. J Am Geriatr Soc 2007;55(8): 1168e1175. Schmader KE. Herpes zoster in the long-term setting. Ann Long Term Care 2004;12:38e42. Centers for Medicare & Medicaid Services. MDS 2.0 public quality indicator and resident reports. Overview. US Department of Health & Human Services 2009:11e16. Hornberger J, Robertus K. Cost-effectiveness of a vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. Ann Intern Med 2006;145(5): 317e325. Pellissier JM, Brisson M, Levin MJ. Evaluation of the cost-effectiveness in the United States of a vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. Vaccine 2007;25(49):8326e8337. Rothberg MB, Virapongse A, Smith KJ. Cost-effectiveness of a vaccine to prevent herpes zoster and postherpetic neuralgia in older adults. Clin Infect Dis 2007;44(10):1280e1288. Patel MS, Gebremariam A, Davis MM. Herpes zoster-related hospitalizations and expenditures before and after introduction of the varicella vaccine in the United States. Infect Control Hosp Epidemiol 2008;29(12):1157e1163. McCullagh P, Nelder JA. Generalized Linear Models. 2nd ed. London: Chapman and Hall; 1989. Agresti A. Categorical Data Analysis. 2nd ed. Hoboken, NJ: John Wiley & Sons; 2002. Stein AN, Britt H, Harrison C, et al. Herpes zoster burden of illness and health care resource utilisation in the Australian population aged 50 years and older. Vaccine 2009;27(4):520e529. Gauthier A, Breuer J, Carrington D, et al. Epidemiology and cost of herpes zoster and post-herpetic neuralgia in the United Kingdom. Epidemiol Infect 2009; 137(1):38e47. Carapetis JR, Russell DM, Curtis N. The burden and cost of hospitalised varicella and zoster in Australian children. Vaccine 2004;23(6):755e761. Schmader KE, Dworkin RH. Natural history and treatment of herpes zoster. J Pain 2008;9(1 Suppl. 1):S3eS9. Schmader K, Gnann JW Jr, Watson CP. The epidemiological, clinical, and pathological rationale for the herpes zoster vaccine. J Infect Dis 2008; 197(Suppl 2):S207eS215.