Burden of herpes zoster and postherpetic neuralgia in Jinan, China: incidence, recurrence, and direct medical costs, 2018–2023, a population-based retrospective cohort study

Abstract

Background

Herpes zoster (HZ) and postherpetic neuralgia (PHN) cause substantial clinical and economic burdens, especially in older adults, yet real-world data from China are limited.

Methods

Using the Jinan Regional Health Information Platform, we conducted a population-based retrospective cohort study (2018–2023) to estimate HZ and PHN incidence (per 1,000 person-years [PY]), recurrence rates, and direct medical costs and factors associated with costs.

Results

The study included 166,869 patients with HZ. The coverage-adjusted incidence of HZ was 6.80 per 1,000 person-years (PY) overall and ranged from 9.68 per 1,000 PY in the 50–59-year age group to a peak of 16.48 per 1,000 PY in the 70–79-year age group. PHN occurred in 14.2% of HZ patients. The overall coverage-adjusted incidence of PHN was 0.94 per 1,000 PY. HZ recurrence occurred in 0.87% of patients, corresponding to an incidence of 4.26 per 1,000 PY. A total of 72.9% of patients were included in the cost analyses. Median annual costs per patient were CNY 155.2 (interquartile range [IQR]: 50.3–533.1) overall, CNY 126.7 (IQR: 44.4–365.6) for outpatients and CNY 6,290.2 (IQR: 4,038.4–9,093.6) for inpatients. Higher costs were associated with older age, PHN, healthcare facility type, more frequent visits, and length of stay.

Conclusions

HZ and PHN impose substantial epidemiological and economic burdens in Jinan, with incidence increasing with age, a persistent risk of recurrence, and costs influenced by demographic and healthcare-related factors.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-13245-2.

Introduction

Herpes zoster (HZ) is a neurocutaneous disease caused by reactivation of latent varicella-zoster virus (VZV) in sensory ganglia [1]. HZ is characterized by a prolonged course, severe pain lasting several weeks, and vesicular eruptions localized to the affected dermatome [1]. The most common complication of HZ is postherpetic neuralgia (PHN), which occurs in 2.6–46.7% of patients and is characterized by persistent pain lasting for months or even years, severely affecting quality of life [2].

The incidence of HZ increases significantly with age. In China, meta-analyses estimate an annual incidence of 4.28 per 1,000 person-years (PY) in the general population, and 11.69 per 1,000 PY among individuals aged 60 years and older [3]. The incidence of HZ also increased between 2000 and 2020 [4]. PHN incidence ranges from 0.15 to 2.1 per 1,000 PY in international settings; elderly populations and immunocompromised individuals have a higher incidence than the general population [2]. HZ recurrence occurs at rates of 1.7–16.6 per 1,000 PY in the general population and 17–55 per 1,000 PY among immunocompromised individuals [5].

HZ was associated with a disease burden of 59.99 disability-adjusted life years per 100,000, which was relatively high among adults aged >50 years [6]. In China, HZ-related hospitalization costs per patient increased from Chinese Yuan (CNY) 3,260 in 2010–2012 to CNY 14,303 in 2017–2018, and outpatient costs increased from CNY 336 to CNY 1,329 [4]. HZ patients with PHN had higher costs than those without PHN, with annual average total expenses exceeding CNY 40,000 and out-of-pocket costs averaging CNY 16,873 [7]. In 2017, the national direct medical cost of HZ was estimated at CNY 9.2 billion [8].

In China, 22.0% of the population (310 million people) was aged 60 years or older in 2024, and this proportion is projected to reach 39.8% by 2050 [9, 10]. With rapid population aging, the health and economic burdens of HZ are expected to continue to increase.

HZ is not a notifiable infectious disease in China. Although existing studies have provided preliminary insight into the epidemiology and economic burden of HZ, most are limited by small sample sizes or are single-center studies, making it difficult to estimate the true population-level burden. Long-term, systematic, population-based surveillance data remain scarce.

Jinan City, the capital of Shandong Province, established the Regional health information platform (RHIP) to integrate healthcare data across most secondary and tertiary hospitals and some primary care institutions. As the platform’s coverage gradually expanded to include a larger proportion of the local population, its representativeness steadily improved. From 2018 onward, the data were sufficiently consistent to support comparable population-based analyses of the burden of HZ.

In contrast to most previous studies in China that relied on small sample sizes or single-center data, our study leveraged large-scale, multi-center real-world data from the Jinan RHIP, enabling a more comprehensive and population-based assessment of the epidemiological and economic burden of HZ and PHN. Using RHIP data, we estimated the incidence of HZ and PHN, the recurrence rate of HZ, and direct medical costs of HZ in Jinan from 2018 to 2023.

Methods

Data sources

We extracted clinical data from Jinan RHIP on all outpatient and inpatient visits with a primary or secondary diagnosis of HZ from January 1, 2018, to December 31, 2023. We obtained the official directory of secondary and tertiary hospitals in Jinan (2018–2023) from the Shandong Provincial and Jinan Municipal Health Commissions. We also extracted citywide, category-level data on the number of healthcare institutions and annual visit volumes for the same period. We screened the list of all secondary and tertiary hospitals in the city, retaining general, Traditional Chinese Medicine, integrated Chinese and Western medicine, dermatology, and ophthalmology hospitals, and excluding specialized institutions not typically involved in HZ diagnosis or treatment (e.g., maternity, pediatric, and dental hospitals), to calculate coverage. Jinan resident population data (2018–2023) were obtained from the information system of the Chinese Center for Disease Control and Prevention.

Case definition and outcome measures

Case definition

HZ cases were identified using International Classification of Diseases, 10th Revision (ICD-10) codes B02.0–B02.9 and corresponding diagnostic terms recorded in Western medicine and Traditional Chinese Medicine fields, including “带状疱疹 (herpes zoster)”, “Ramsay-Hunt 综合征 (Ramsay-Hunt syndrome)”, as well as Traditional Chinese Medicine diagnoses such as “蛇串疮 (snake-like clustered sores)”, “蛇丹 (snake-related rash)” and “缠腰火丹 (waist-encircling fire rash)”. PHN cases were identified among patients with HZ based on diagnostic terms explicitly indicating postherpetic neuralgia, including “带状疱疹后神经痛 (postherpetic neuralgia)”. The full list is provided in Supplementary Table S1. Given that diagnostic terms do not encode disease duration, we conducted a sensitivity analysis using a time-anchored PHN definition, in which PHN was defined as the presence of a PHN-related diagnostic term recorded ≥ 90 days after the index HZ diagnosis. A flowchart illustrating PHN classification is provided in Supplementary Figure S2.

For each patient, the first recorded HZ-related visit during the study period was defined as the initial HZ episode (index visit). HZ recurrence was defined as two diagnoses in the same patient separated by an interval of ≥ 180 days [5]. As a sensitivity analysis, recurrence was alternatively defined using an interval of ≥365 days. Both first-episode and recurrent cases were included in the analysis. PHN was considered a single event per patient, and therefore PHN recurrence was not included in the analysis.

Epidemiological measures

Unadjusted incidence per 1,000 PY was calculated as (number of new HZ cases in RHIP in a given year)÷(resident population in the same year)×1,000; recurrence rate per 1,000 PY was calculated as (number of recurrent episodes)÷(total person-years of follow-up)×1,000. Follow-up time was defined as the interval from the first HZ diagnosis to December 31, 2023 (the end of the study period), or to the date of death, for subjects who died during the study period.

We applied an indirect, visit-weighted coverage adjustment to account for incomplete inclusion of healthcare institutions in the RHIP. For each calendar year, category-level total visit counts for all diseases by facility type (tertiary hospitals, secondary hospitals, and community health centers) were used to calculate the proportion of total citywide visits attributable to each facility type, with total visits across all healthcare institutions as the denominator. These proportions were then weighted by the corresponding RHIP coverage proportions at each facility type, and the sum of the weighted proportions was defined as the annual coverage adjustment factor (k). Coverage-adjusted incidence rates were calculated by dividing unadjusted incidence rates by k. PHN incidence rates were adjusted using the same approach. Details of the calculation of k are described in Supplementary Table S2.

Direct medical cost analysis

Direct medical costs were defined as quantifiable costs for HZ-related care, including five major categories: surgical costs, diagnostic fees (e.g., testing, radiology, and laboratory fees), medication costs (Western medicines, Chinese herbal medicines, proprietary Chinese medicines, blood products and albumin products), non-surgical treatment costs (e.g., nursing fees, general treatment procedure fees, rehabilitation fees, and consumable costs), and other costs (e.g., accompanying bed fees and service fees). All cost information was obtained directly from the itemized billing records in the Jinan RHIP. Costs were extracted according to the predefined billing categories in the system and then grouped into the five cost categories described above. For the direct medical cost analyses, we included only patients whose visits had complete cost information and for whom HZ was recorded as the primary diagnosis. All costs were adjusted to 2023 values using the Chinese consumer price index.

Based on the presence of hospitalization records, patients were classified into outpatient or inpatient groups, and direct medical costs were analyzed separately. Inpatient cases were defined as patients with at least one hospitalization record and were excluded from the outpatient cost analyses. Multivariable models were constructed to identify factors associated with cost. Annualized per-patient costs were calculated as (total HZ-related cost per patient) ÷ (number of years spanned by visits). The number of years spanned by visits was the interval between first and last HZ-related visits; for patients with only one visit, the number of years spanned by visits was defined as 1 year. Per-episode costs were calculated by summing HZ-related direct medical costs across all encounters within an episode. Episodes were defined using a 180-day washout interval, consistent with the recurrence definition.

Statistical analysis

Data analyses were performed using R software (version 4.2.1). Age was summarized using medians with interquartile ranges (IQRs). Annual per-patient costs were summarized as medians with interquartile ranges, and mean values with standard deviations (mean ± SD) were also reported for descriptive purposes.

Categorical variables were presented as counts and percentages. Between-group comparisons were performed using the Mann-Whitney U test or Kruskal-Wallis H test for continuous variables and the chi-square (χ²) test for categorical variables. The 95% confidence intervals (CIs) for incidence and recurrence rates were estimated using Poisson approximation. Annual trends in incidence were assessed using Poisson regression models with coverage-adjusted case counts as the outcome and PY as the offset. To assess the robustness of temporal trend estimates to changes in population coverage, sensitivity analyses were conducted by restricting the trend models to calendar periods with relatively stable coverage (2019–2021 and 2022–2023), applying the same Poisson regression models. Recurrence was further characterized using the Kaplan-Meier method; time to recurrence was defined from the index visit to the first recurrent HZ episode (≥ 180-day definition), with censoring at death or end of follow-up (December 31, 2023). Factors associated with direct medical costs were analyzed using generalized linear models (GLMs). Given the right-skewed distribution of costs, a Gamma distribution with a log link function was used in cost analyses. Results are presented as exponentiated regression coefficients [exp(β)] with 95% CIs. In the GLMs, costs were analyzed at the patient level, with each patient contributing a single observation representing the sum of all costs from their HZ-related visits. Two-sided P values < 0.05 were considered statistically significant.

Results

Baseline study population characteristics

During the study period of 2018–2023, there were 277,495 HZ-related visits identifiable in the Jinan RHIP. After excluding records without age or sex information and deduplicating records by anonymized patient ID, data on 166,869 HZ patients remained and were included in the study population; 77.8% of patients had only one recorded visit. During the follow-up period, 168,465 HZ episodes occurred, comprising 166,869 initial episodes and 1,596 recurrences. For the cost analyses, records with non-primary HZ diagnoses or incomplete cost information were excluded, leaving 121,591 patients (72.9% of the HZ study population) in the cost analyses (Supplementary Figure S1).

Patient age and sex distributions are shown in Supplementary Figure S3. The median age of included patients was 58 (46–68) years. Patients with PHN were significantly older than patients without PHN (63 vs. 58 years, P < 0.05), and the proportion of males was higher among patients with PHN than among those without PHN (48.7% vs. 47.3%, P < 0.05). Compared with the full cohort, the cost-analysis cohort had similar age and sex distributions (median age 58 years in both; females 52.7% vs. 52.5%), but lower proportions of care at secondary- or higher-level hospitals (57.8% vs. 65.5%) and inpatient care (8.9% vs. 13.7%), and a higher proportion of PHN (15.8% vs. 14.2%) (Supplementary Table S3), with PHN counted once per patient.

HZ and PHN incidence rates

During the six-year study period, the unadjusted annual incidence of HZ ranged from 1.94 to 5.27 per 1,000 PY and the unadjusted annual incidence of PHN ranged from 0.19 to 0.89 per 1,000 PY. After adjustment for RHIP coverage, the estimated incidence of HZ and PHN increased substantially. The coverage-adjusted HZ incidence ranged from 4.34 to 10.33 per 1,000 PY, and the coverage-adjusted PHN incidence ranged from 0.50 to 1.73 per 1,000 PY. Overall, both outcomes showed statistically significant increasing temporal trends over the study period (P < 0.05) (Table 1). However, when restricting the trend models to shorter periods with relatively stable coverage (2019–2021 and 2022–2023), the estimated year effects were not consistently positive, with negative year coefficients observed in some sub-period analyses.

Table 1.

Annual unadjusted and coverage-adjusted incidence of herpes zoster and postherpetic neuralgia, 2018–2023

Year	HZ episodes	Unadjusted HZ incidence (per 1,000 PY)	Coverage-adjusted HZ incidence (per 1,000 PY)	PHN cases (n)	Unadjusted PHN incidence (per 1,000 PY)	Coverage-adjusted PHN incidence (per 1,000 PY)
2018	16,521	2.26	5.96 (5.91–6.02)	1,405	0.19	0.50 (0.49–0.52)
2019	20,279	2.29	5.33 (5.28–5.38)	2,382	0.27	0.65 (0.63–0.66)
2020	17,316	1.94	4.34 (4.30–4.39)	2,191	0.25	0.55 (0.53–0.57)
2021	18,910	2.05	4.62 (4.58–4.67)	2,736	0.30	0.67 (0.65–0.69)
2022	48,954	5.27	10.33 (10.26–10.39)	8,221	0.89	1.73 (1.71–1.76)
2023	46,485	4.94	9.83 (9.76–9.89)	6,779	0.72	1.38 (1.35–1.40)
Total	168,465	3.18	6.80 (6.78–6.83)	23,714	0.48	0.94 (0.93–0.95)

Abbreviations: Values in parentheses are 95% CIs

Coverage-adjusted incidences of HZ and PHN generally increased with age. Using 0–9-year-olds as the reference, the relative risk (RR) of HZ increased from 5.40 (95% CI: 5.03–5.80) among 10–19-year-olds to an age-group maximum of 105.93 (95% CI: 99.21–113.09) among 70–79-year-olds; among the oldest age group of individuals aged ≥ 80 years, RR was 102.41 (95% CI: 95.85–109.43). Age-specific differences were greater for PHN, with an RR for PHN of 393.65 (95% CI: 295.71–524.02) among individuals ≥ 80 years of age. Females had higher RRs for both HZ and PHN (Table 2). Age-specific trends in HZ and PHN incidence across different study years are shown in Supplementary Figure S4.

Table 2.

Coverage-adjusted incidence and relative risk of herpes zoster and postherpetic neuralgia by age group and sex

Group	HZ Adjusted Incidence	HZ RR	PHN Adjusted Incidence	PHN RR
Age group
0–9	0.16 (0.15–0.17)	Reference	0.01 (0.01–0.01)	Reference
10–19	0.84 (0.82–0.87)	5.40 (5.03–5.80)	0.05 (0.05–0.06)	6.36 (4.67–8.66)
20–29	2.51 (2.48–2.56)	16.16 (15.12–17.27)	0.15 (0.14–0.16)	18.93 (14.16–25.32)
30–39	4.87 (4.82–4.92)	31.31 (29.32–33.44)	0.41 (0.40–0.43)	50.85 (38.18–67.72)
40–49	6.18 (6.12–6.23)	39.69 (37.17–42.38)	0.71 (0.69–0.72)	87.06 (65.42–115.86)
50–59	9.68 (9.61–9.74)	62.17 (58.24–66.37)	1.24 (1.22–1.27)	153.36 (115.30–203.98)
60–69	15.29 (15.19–15.39)	98.23 (92.02–104.85)	2.45 (2.41–2.49)	301.97 (227.08–401.57)
70–79	16.48 (16.35–16.62)	105.93 (99.21–113.09)	3.02 (2.97–3.08)	373.10 (280.51–496.26)
≥80	15.94 (15.73–16.15)	102.41 (95.85–109.43)	3.19 (3.10–3.28)	393.65 (295.71–524.02)
Sex
Male	6.38 (6.35–6.41)	Reference	0.90 (0.89–0.92)	Reference
Female	7.24 (7.20–7.27)	1.13 (1.13–1.14)	0.98 (0.97–0.99)	1.09 (1.07–1.10)

Abbreviations: Values in parentheses are 95% CIs. Incidence is adjusted for RHIP-inclusion coverage

Using a stricter, time-anchored definition requiring PHN diagnoses to occur ≥ 90 days after the index HZ episode, 1,110 of 166,869 patients (0.67%) met the PHN criteria (Supplementary Figure S2). Under this definition, the overall coverage-adjusted PHN incidence was 0.047 per 1,000 PY. Age- and sex-specific PHN incidence estimates based on the time-anchored definition are shown in Supplementary Table S4.

HZ recurrence

The study cohort included 166,869 patients with a median follow-up of 1.70 years. During follow-up, 1,455 patients experienced recurrence, accounting for 0.87% of all HZ patients, with a total of 1,596 recurrent HZ episodes recorded. The overall recurrence rate was 4.26 per 1,000 PY (95% CI: 4.05–4.48). Age-stratified analysis showed that recurrence risk increased with age, from 0.75 per 1,000 PY in the 0–9-year age group to 8.24 per 1,000 PY in the ≥ 80-year age group (Table 3). Using a stricter recurrence threshold of ≥ 365 days, the recurrence rate was 2.23 per 1,000 PY (835 events), compared with 4.26 per 1,000 PY under the ≥ 180-day definition.

Table 3.

Recurrence rate of herpes zoster by age group and sex

Group	HZ cases	Follow-up person-years	Recurrent episodes	Recurrence rate (per 1,000 PY)	95% CI
Age group
0–9	398	1,340	1	0.75	0.02–4.16
10–19	1926	5,774	2	0.35	0.04–1.25
20–29	8093	21,090	39	1.85	1.31–2.53
30–39	18,096	42,440	164	3.86	3.30–4.50
40–49	22,086	51,030	264	5.17	4.57–5.84
50–59	37,720	82,878	330	3.98	3.56–4.44
60–69	42,664	93,824	330	3.52	3.15–3.92
70–79	25,644	53,595	280	5.22	4.63–5.87
≥ 80	10,242	22,568	186	8.24	7.10–9.52
Sex
Male	79,289	177,513	755	4.27	3.98–4.57
Female	87,580	197,025	841	4.25	3.96–4.57
Total	166,869	374,538	1,596	4.26	4.05–4.48

The cumulative recurrence probability increased most rapidly during the first 2–3 years after the index visit and then increased more slowly thereafter (Fig. 1). The cumulative recurrence probability was 0.45% (95% CI: 0.41%–0.49%) at 1 year, 1.34% (95% CI: 1.26%–1.41%) at 3 years, and 1.68% (95% CI: 1.58%–1.78%) at 5 years. Cumulative recurrence did not differ significantly by sex (log-rank P = 0.76).

Fig. 1.

Kaplan-Meier estimates of cumulative recurrence probability following the index visit. The shaded area indicates the 95% CI

Direct medical costs

A total of 121,591 patients were included in the cost analysis, with direct medical costs of CNY 123,609,859. The mean direct medical cost per patient was CNY 1,016.6 ± 3,758.9.

Outpatients accounted for 110,715 (91.1%) patients but CNY 38,672,301 (31.3%) of total direct medical costs, whereas inpatients accounted for 10,876 (8.9%) patients but CNY 84,937,558 (68.7%) of total costs. Among the 19,238 (15.8%) patients with PHN, medical costs totaled CNY 25,919,227, representing 21.0% of all costs. Medication costs accounted for the largest proportion of costs (58.1%), followed by non-surgical treatment costs (26.4%). Diagnostic and surgical fees accounted for a relatively small proportion (3.2%) of costs. For inpatients, non-surgical treatment costs accounted for the largest proportion of costs (38.7%), followed by medication costs (33.9%). In contrast, medication costs accounted for 78.5% of outpatient costs.

The annual median direct medical cost per HZ patient was CNY 155.2 (50.3–533.1) overall. The corresponding annual median costs were CNY 126.7 (44.4–365.6) for outpatients and CNY 6,290.2 (4,038.4–9,093.6) for inpatients. In outpatients and inpatients, costs differed significantly by age group, PHN status, healthcare facility type, and number of visits (all P < 0.05; Table 4). The episode-level cost analysis included 121,591 initial herpes zoster episodes and 799 recurrent episodes. Episode-level direct medical cost estimates are presented in Supplementary Table S5. On a per-episode basis, episodes without PHN had a mean cost of CNY 945.8 and a median cost of CNY 144.5, whereas episodes with PHN had a mean cost of CNY 1,325.6 and a median cost of CNY 230.6.

Table 4.

Median costs of outpatients and inpatients with herpes zoster by patient characteristics

Characteristic	Outpatients				Inpatients
	N	PHN, n (%)	Median cost, CNY		N	PHN, n (%)	Median cost, CNY
Sex
Female	58,399	9,090 (15.6)	122.8 (43.0–361.3)		5,729	945 (16.5)	6,235.3 (4,028.7–8,915.7)
Male	52,316	8,356 (16.0)	131.0 (45.9–371.5)		5,147	847 (16.5)	6,346.6 (4,048.6–9,271.6)
Age group (years)
<40	20,372	1,976 (9.7)	142.4 (53.6–357.1)		1,041	88 (8.5)	5,727.9 (3,736.3–7,733.1)
40–49	15,597	2,181 (14.0)	139.6 (48.1–383.3)		1,044	100 (9.6)	5,817.9 (3,780.0–8,109.6)
50–59	25,326	3,864 (15.3)	123.9 (44.2–359.4)		2,698	404 (15.0)	5,827.8 (3,623.4–8,438.2)
60–69	28,161	5,012 (17.8)	115.3 (40.1–360.1)		3,116	605 (19.4)	6,729.8 (4,304.5–9,664.7)
≥70	21,259	4,413 (20.8)	119.5 (40.1–382.0)		2,977	595 (20.0)	6,701.5 (4,350.2–10,066.6)
PHN status
No	93,269	0 (0.0)	119.3 (43.0–337.3)		9,084	0 (0.0)	6,058.1 (3,921.0–8,807.1)
Yes	17,446	17,446 (100.0)	185.0 (51.9–548.2)		1,792	1,792 (100.0)	7,442.9 (4,951.7–10,560.2)
Healthcare facility type
Primary healthcare institutions	50,729	7,171 (14.1)	66.7 (29.9–133.7)		619	60 (9.7)	3,743.4 (2,503.2–5,555.3)
Secondary or higher-level hospitals	59,986	10,275 (17.1)	266.0 (96.4–656.7)		10,257	1,732 (16.9)	6,460.0 (4,193.4–9,251.8)
Number of visits
1	88,599	12,264 (13.8)	90.1 (32.9–214.5)		8,728	1,241 (14.2)	5,807.5 (3,815.3–8,437.7)
≥ 2	22,116	5,182 (23.4)	735.0 (344.1–1,349.9)		2,148	551 (25.7)	8,430.8 (5,792.5–11,732.3)
Length of stay (days)
≤ 7					4485	599 (13.4)	4,215.8 (2,959.8–5,888.1)
8–13					4816	817 (17.0)	7,214.0 (5,277.2–9,189.8)
≥ 14					1,575	376 (23.9)	12,573.9 (9,757.6–17,352.0)
Total	110,715	17,446 (15.8)	126.7 (44.4–365.6)		10,876	1,792 (16.5)	6,290.2 (4,038.4–9,093.6)

Abbreviations: Values in parentheses are presented as IQRs; PHN status is shown for completeness; PHN proportions in other strata are calculated within each category; Number of visits refers to HZ-related visits defined by herpes zoster as the primary diagnosis

Gamma-log GLM regression analysis showed that the deviance-to-degrees-of-freedom (deviance/df) for the outpatient model was 1.24, and for the inpatient model it was 0.29. Among outpatients, males had higher costs than females (cost ratio 1.04, 95% CI: 1.03–1.06, P < 0.001). Costs increased significantly with age, particularly in patients aged ≥ 60 years compared with those < 40 years (P < 0.001). PHN was associated with significantly higher outpatient costs compared with non-PHN status (P < 0.001). Visits to secondary or higher-level hospitals and multiple visits were associated with significantly higher costs (P < 0.001), with healthcare facility type and number of visits having the greatest association with outpatient costs.

Among inpatients, age ≥ 60 years, hospitalization at secondary or higher-level hospitals, longer length of stay (≥ 8 days), and PHN occurrence were all significantly associated with increased costs (all P < 0.001), whereas sex was not significantly associated (P = 0.069). Length of stay had the greatest impact on inpatient costs; patients hospitalized for ≥ 14 days incurred 3.258 times the cost of those hospitalized ≤ 7 days (Table 5).

Table 5.

Multivariable analysis of outpatient and inpatient costs for herpes zoster patients

Characteristic	Outpatients			Inpatients
	exp(β) (95% CI)	P value		exp(β) (95% CI)	P value
Sex
Female	Ref			Ref
Male	1.042 (1.029–1.056)	< 0.001		1.028 (0.998–1.059)	0.069
Age group (years)
<40	Ref			Ref
40–49	1.115 (1.089–1.141)	< 0.001		1.012 (0.946–1.082)	0.734
50–59	1.113 (1.090–1.136)	< 0.001		1.025 (0.969–1.084)	0.384
60–69	1.192 (1.169–1.217)	< 0.001		1.154 (1.093–1.219)	< 0.001
≥70	1.312 (1.285–1.340)	< 0.001		1.167 (1.104–1.233)	< 0.001
PHN status
No	Ref			Ref
Yes	1.407 (1.385–1.430)	< 0.001		1.090 (1.050–1.131)	< 0.001
Healthcare facility type
Primary healthcare institutions	Ref			Ref
Secondary or higher-level hospitals	2.210 (2.173–2.249)	< 0.001		1.716 (1.613–1.827)	< 0.001
Number of visits
1	Ref			Ref
≥ 2	5.698 (5.607–5.790)	< 0.001		1.065 (1.030–1.101)	< 0.001
Length of stay (days)
≤ 7				Ref
8–13				1.597 (1.544–1.652)	< 0.001
≥ 14				3.258 (3.115–3.407)	< 0.001

Abbreviations: Number of visits refers to HZ-related visits with herpes zoster as the primary diagnosis

Discussion

Our large, retrospective cohort study based on the Jinan City Regional Health Information Platform showed that during 2018–2023 the annual all-ages coverage-adjusted incidence of HZ was 6.80 per 1,000 PY and increased with age, ranging from 9.68 to 16.48 episodes per 1,000 PY among individuals 50 years and older; 14.2% of HZ patients had PHN, and the annual coverage-adjusted incidence per 1,000 PY of PHN ranged from 0.50 to 1.73. The annual recurrence rate of HZ was 4.26 per 1,000 PY, and recurrence risk increased with time from initial HZ episode. Coverage-adjusted incidence was higher in women than men for both HZ and PHN. Annual median costs per patient were 126.7 CNY for outpatients and 6,290.2 CNY for inpatients, and were higher for patients with PHN. HZ and PHN impose substantial health and economic burdens in Jinan, China.

Our age-specific incidence estimates generally fall within the range reported in prior studies in China and elsewhere [4, 11–14]. A community-based survey conducted in Laiwu District (Jinan) reported an incidence of 12.37–15.11 per 1,000 PY in individuals aged ≥ 60 years in 2018, which was comparable to the findings of the present study [15]. A meta-analysis conducted in China showed that the overall incidence of HZ was 6.28 per 1,000 PY (95% CI: 5.42–7.15), which is comparable to our findings [3]. However, the incidence among individuals aged ≥ 50 years was lower than in our study, highlighting potential regional differences and variations in population demographics that may influence the disease burden [3]. For instance, a study in Beijing in 2015 reported more than a threefold variation in HZ incidence across three districts, with rates of 16.2, 4.0, and 5.7 per 1,000 PY in Xicheng, Changping, and Miyun, respectively [16]. Globally, the reported incidence of HZ in individuals ≥ 50 years ranges from 5.23 to 10.9 per 1,000 PY, with estimates in the Asia-Pacific region spanning 2.9 to 19.5 per 1,000 population [12]. The upper range of our estimates for Jinan (16.48 per 1,000 PY) aligns with the higher end of this international spectrum. A recent population-based study in Sweden also reported a lower overall incidence of 4.2 per 1,000 PY [13]. Taken together, these findings indicate substantial geographic heterogeneity in HZ incidence across regions and within cities. The relatively high coverage-adjusted incidence of HZ that we found in Jinan may be partially explained by the high concentration of medical resources, leading to patient referrals and healthcare-seeking from surrounding regions, as well as differences in population coverage and case ascertainment over time.

Importantly, our year-to-year pattern was clearly non-monotonic and should be interpreted in the context of the COVID-19 pandemic and changes in healthcare utilization and ascertainment over time (Table 1; Figure S4). Specifically, the coverage-adjusted HZ incidence was lower in 2020–2021 than in 2018–2019, and then rebounded markedly in 2022–2023 (Table 1; Figure S4). A plausible explanation is pandemic-related service disruptions and altered care-seeking/diagnostic practices during 2020–2021, followed by restoration of healthcare access and a rebound in healthcare utilization after major pandemic phases [17]. Notably, the PHN incidence in 2020–2021 was similar to that in 2018–2019 (Table 1), suggesting comparatively stable ascertainment of PHN diagnoses across the pandemic years. In addition, community health centers were incorporated into RHIP only in 2022–2023. The expanded capture of visits in these settings could have increased case ascertainment among older adults, who may be more likely to seek primary care for convenience and chronic disease management. This may have modestly biased age-specific incidence estimates upward in the later period.

The proportion of HZ patients with a diagnosis of PHN (14.2%) that we found is similar to that reported in a longitudinal cohort study in the United States (12.8%) [18], a systematic review and meta-analysis from Latin America (12.9–21.8%) [19], and a pooled risk estimate from a meta-analysis from China (12.6%, 95% CI: 10.1–15.1) [3]. Our estimate was higher than that reported in a systematic review from the Asia-Pacific region (4.0–12.4%) [11]. Variation in the reported PHN burden may stem from methodological differences, including definitions, study design, and population characteristics, as well as differences in comorbidity prevalence and immune status. Using a time-anchored (≥ 90 days) strict definition, the estimated coverage-adjusted PHN incidence was lower than the estimate based solely on diagnosis terms. This pattern suggests that diagnosis-based definitions may overestimate PHN by capturing acute or subacute HZ-related pain, whereas time-anchored definitions may underestimate PHN due to limited follow-up, end-of-study truncation, and care received outside the captured system. Together, these two approaches help contextualize the plausible range of the PHN burden. It should be noted that in the primary analysis, PHN was identified based on diagnostic terms recorded in physicians’ diagnostic records, which do not capture the specific time point at which clinicians determined that pain had met a time-based PHN definition. Therefore, the primary analysis reflects PHN diagnoses as documented in routine clinical practice, while the time-anchored sensitivity analysis was used to evaluate PHN estimates under an explicit temporal constraint.

In our study, the proportion of patients with HZ recurrence (0.87%) was lower than previously reported (1.2%–9.6%), while the recurrence rate (4.26 per 1,000 PY) was within, but at the lower end of, the reported range (1.7–16.6 per 1,000 PY) [5]. The lower recurrence rate that we found may be due in part to the relatively short follow-up time in our study (median, 1.7 years), which primarily captures early recurrences and may underestimate longer-term risk. Longer-term studies have shown that recurrence risk accumulates over time, with a cumulative risk over 10 years exceeding 10% [5]. In our Kaplan-Meier analysis, recurrence accumulated most rapidly in the first few years after the index episode and then the cumulative curve gradually flattened, indicating that recurrence accumulated more slowly later in follow-up within the observed window. This time pattern differed from that reported in an Australian population-based study, which observed a reduced recurrence incidence in the first 12 months after the initial episode, followed by a relatively constant annual recurrence incidence thereafter—an evolution that would translate into a more nearly linear cumulative curve over longer follow-up [20]. Moreover, some cohorts have reported an approximately linear increase in cumulative recurrence across long-term follow-up [21]. Differences in follow-up duration and censoring, washout intervals used to define recurrence, case ascertainment strategies, and healthcare-seeking behavior may therefore contribute to the apparent deceleration observed in our data. These findings suggest that while recurrence risk persists over time, short- to medium-term recurrence rates may remain relatively stable in certain populations. Nevertheless, given that the risk of HZ recurrence is not confined to the short term and increases markedly with age [20–22], recurrence should be considered in prevention planning for older adults. Estimates of HZ recurrence differ across studies in part because varying time intervals are used to define recurrence, reflecting the absence of a standardized definition in observational research [5]. Our sensitivity analysis using a stricter recurrence threshold was conducted to assess how recurrence estimates change with the choice of definition, rather than to redefine recurrence itself. Finally, we observed no sex difference in cumulative recurrence in our cohort (log-rank P = 0.76). Although several prior cohorts have reported higher recurrence rates among women [20–22], other population-based evidence, such as a population-based study from China, has not demonstrated a comparable sex disparity [23]. The reasons for these inconsistent findings remain unclear and warrant further investigation.

Our study used annual per-patient costs as a primary metric for HZ and PHN health care costs. Most studies in China report costs on a per-case basis [24], and international studies tend to use episode-related or per-case costs. For example, a Canadian study reported the average per-case costs of HZ and PHN and the total annual national costs, rather than annual per-patient costs [25]. Similarly, studies conducted in Germany [26], Italy [27], Spain [28] and Singapore [29] reported costs on a per-episode or per-case basis. However, some analyses using insurance claims data use annual per-patient costs. An example is a U.S. study that assessed the annual economic burden of HZ and PHN patients [30]. Among patients with a single healthcare visit, annual per-patient costs were consistent with per-case costs. In our study, the annual per-patient outpatient costs for single-visit patients were lower than the per-case costs reported in certain regions of China [31, 32]. For patients with multiple visits, the median annual outpatient cost for patients with multiple visits in our study was 735.0 CNY, which was lower than the corresponding estimates reported for multiple-visit outpatients in Laiwu District of Jinan (1,480 CNY) [15] and Beijing (1,000 CNY) [32]. Since per-patient and per-episode costs will differ when a subject has more than one episode, per-patient studies with longer cumulative follow-up time are likely to give higher costs than per-episode studies and may be more relevant for patient-focused analyses. To improve comparability with studies reporting episode-based costs, we additionally summarized HZ-related direct medical costs on a per-episode basis. These episode-based estimates complement the primary annual per-patient costs by providing an alternative costing unit and additional context for interpretation and cross-study comparison.

Our stratified analyses showed that frequency of visits, hospital level, hospitalization status, and length of hospital stay were all significantly associated with costs, consistent with other studies. For example, a study in Beijing found that costs in tertiary hospitals were higher than those in secondary hospitals [32]. Research from Ningbo and Laiwu District of Jinan showed that hospitalized patients incurred substantially higher costs than outpatients [15, 22], while an investigation in Hunan found that longer hospital stays, PHN, and immunosuppression were significantly associated with cost [33]. Our study found that inpatient costs were mainly driven by non-surgical treatment and consumable costs, whereas outpatient costs were driven by medication costs. These findings suggest that structural differences across healthcare settings should be considered in cost containment strategies and cost-effectiveness evaluations.

PHN increased the economic burden of HZ. Patients in our study with PHN incurred significantly higher costs than patients without PHN, a finding consistent with existing evidence. In China, studies conducted in rural areas of several provinces and in urban communities of Shanghai reported that costs were significantly associated with hospitalization and PHN [34]. Research from Beijing showed that longer PHN duration was linked to substantially higher costs [32]. International data also support these observations: in Canada, although PHN patients represented fewer than 10% of cases, patients with PHN accounted for nearly 35% of total HZ costs [25]. A study in the United Kingdom estimated that PHN-related expenses constituted up to 72% of overall economic HZ burden [35]. Our cost analyses were restricted to patients for whom we had complete cost information—about three-fourths of the cohort. Patients with PHN accounted for 15.8% of the cost-study cohort, higher than the 14.2% PHN rate in the overall cohort. Patients with PHN incurred 21.0% of the direct medical costs in our study—a smaller relative difference than reported in some international studies. This more modest cost share does not indicate a lower economic burden of PHN, but rather reflects differences in healthcare utilization patterns between patients with and without PHN. As shown by the multivariable analyses, medical costs were influenced by multiple healthcare utilization characteristics in addition to PHN status, including hospitalization, type of healthcare facility, and number of HZ-related visits; therefore, comparisons based on the proportion of total HZ costs attributable to PHN mainly reflect the overall direction of disease burden, whereas the exact numerical values are not directly comparable across studies. Regarding model fit, the deviance/df for the inpatient model was 0.29. This relatively low value may be related to the strong correlation between inpatient costs and length of stay, while other potential sources of variability may not be fully captured. Future analyses could consider incorporating additional covariates to further explore remaining unexplained cost variability. Because costs were summarized based on those observed within the available follow-up period, our estimates do not represent cumulative multi-year costs; thus, the long-term overall economic burden associated with persistent PHN may not be fully reflected. Nevertheless, PHN remains an important driver of the economic burden of HZ and should be modeled as an independent health state in economic evaluations. Early antiviral treatment within 72 h of HZ onset and prevention with vaccination are both important for reducing the risk of PHN and alleviating the overall disease burden [36].

Strengths of our study include the large dataset and use of a regional health platform that includes diagnostic and complete healthcare cost data, with patient records linked by unique patient identifiers. Use of a maturing regional health data platform makes our study reproducible with methods that can be used in future regional and national comparisons.

Our study has limitations. Rate calculations may be influenced by the breadth of inclusion of the population covered in RHIP and completeness of diagnostic and cost records. We adjusted for incomplete RHIP coverage using an indirect, visit-weighted coverage adjustment factor (k). This approach assumes that facility-level care-seeking patterns for HZ mirror those observed across all diseases and that healthcare-seeking behavior is similar between covered and non-covered populations. If these assumptions do not fully hold, the resulting incidence estimates may remain subject to residual bias. In addition, because community health centers were incorporated into the coverage adjustment only in 2022–2023, any shift in care-seeking toward primary care during these years could differentially affect the adjustment and potentially lead to an overestimation of the coverage-adjusted incidence in 2022–2023. As our study period also overlapped with the COVID-19 pandemic, year-to-year comparisons—particularly between 2020 and 2021 and 2022–2023—should be interpreted cautiously given evolving platform coverage and pandemic-related disruptions [17, 37]. In the cost analyses, we excluded patients with any visit lacking complete cost information or with a non-HZ primary diagnosis. Given that the cost-analysis cohort included fewer inpatients and fewer encounters in secondary- or higher-level hospitals than the full cohort, overall cost levels may be underestimated if higher-cost encounters were disproportionately excluded. In addition to these cost-cohort restrictions, the RHIP platform could not fully capture patient encounters across medical institutions within and outside the regional data platform, particularly in township hospitals or village clinics, which may lead to incomplete ascertainment of healthcare utilization. Consequently, recurrence rates could be underestimated, and total healthcare utilization and costs may also be underestimated. Moreover, because recurrence estimates depend on follow-up duration, relatively short follow-up will further underestimate longer-term recurrence risk. In addition, several data limitations should be noted. Lack of data on important confounders, such as chronic disease status, precludes analyses by meaningful subgroups. RHIP only includes direct medical costs, precluding economic analysis from a societal perspective. The results are specific to Jinan and do not necessarily generalize to other regions or nationwide.

Conclusions

Our study found that the incidence of herpes zoster and postherpetic neuralgia was higher in women than in men and increased with advancing age and over calendar years. Although the overall risk of HZ recurrence was relatively low, risk persisted over time and increased with age, highlighting the need for sustained preventive strategies. The direct medical costs of HZ were modest overall but had substantial variation by healthcare utilization patterns, type of healthcare facility, hospitalization, and presence of PHN.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CI

Confidence interval

CNY

Chinese Yuan

exp(β)

Exponentiated regression coefficient (Cost Ratio)

GLM

Generalized linear model

HZ

Herpes zoster

ICD-10

International Classification of Diseases, 10th Revision

IQR

Interquartile range

PHN

Postherpetic neuralgia

PY

Person-years

RHIP

Regional health information platform

RR

Relative risk

SD

Standard deviation

VZV

Varicella-zoster virus

Author contributions

FW and CL were responsible for the conception and design of the study. CL and SM contributed to data acquisition. ZZ, XW, LT, SL, QL, XL, and ZY were responsible for data analysis and interpretation. ZZ and XW drafted the manuscript, and FW critically revised it for important intellectual content. All authors approved the final version to be submitted for publication.

Funding

This work was supported by the Beijing Natural Science Foundation-Changping Innovation Joint Fund (grant No. L244078). The funding body had no role in the study design, data collection, analysis, interpretation, writing of the manuscript, or the decision to submit the article for publication.

Data availability

The datasets analysed during the current study are not publicly available due to privacy restrictions and institutional data protection policies, but are available from the corresponding authors on reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the principles of the Declaration of Helsinki. The study protocol was reviewed and approved by the Institutional Review Board (Ethics Review Committee) of the Chinese Center for Disease Control and Prevention (Chinese Academy of Preventive Medicine) (Approval No. 202440-01). The requirement for individual informed consent to participate was reviewed and waived by the same Institutional Review Board (Ethics Review Committee) of the Chinese Center for Disease Control and Prevention.

Consent for publication

Not applicable.

Generative AI and AI-assisted technologies in the manuscript preparation process

During the preparation of this work the authors used ChatGPT in order to improve the language. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Competing interests

The authors declare no competing interests.