Herpes zoster is a reactivation of the varicella-zoster virus, which causes chickenpox. Following chickenpox, which usually occurs early in life, the virus lays dormant in a posterior nerve root ganglion. Reactivation is characterized by pain and reddening of the skin, with vesicles in the distribution of the affected nerve root. Unless secondary infection occurs, the skin lesions usually resolve within a week and leave small scars. Although the pain typically subsides with the rash, it can be severe and persist for many weeks or months . Herpes zoster occurs 10 times more often in HIV-infected persons than in HIV-negative persons [2,3] and there is a higher risk of recurrent episodes [2–5]. Herpes zoster is rare in young immunocompetent adults and in sub-Saharan Africa it has been found to be a strong predictor of HIV-1 infection in young adults [5–8] and an indicator of early mortality . Although, there have been many studies from industrialized countries documenting the disease presentation in HIV-infected participants and comparing this with uninfected participants, there have been few studies from sub-Saharan Africa and these have all been from referral centres [5,8].
This study describes the rates of herpes zoster and compares its clinical features in HIV-positive participants and HIV-negative controls enrolled in a prospective population-based cohort study in rural Uganda. In the HIV-positive group, the incidence of herpes zoster is reported from seroconversion, any relation between an episode of herpes zoster and a more rapid progression to death is sought and the prognostic value of CD4 cell counts and other characteristics on incidence is examined.
The study population
The HIV-1 clinical cohort was established in 1990 and recruited participants from a large general population to study the dynamics of HIV-1 infection in rural Southwest Uganda . A random sample of the HIV-positive individuals (prevalent cases) found during the initial survey of the general population cohort, all seroconverters detected during subsequent rounds (incident cases) and age-stratified HIV-negative controls were invited to join the clinical cohort. The estimated date of seroconversion for the incident cases was taken as midway between the last HIV-negative and first HIV-positive result. The clinical cohort has been described in detail elsewhere . At enrolment, all participants were asked about their previous medical history, including if they had ever had herpes zoster (ekisipi) and if so, in which year(s). As part of their full physical examination, the presence and distribution of zoster scarring were noted. An anatomical distribution chart of dermatomes was used to estimate the affected area of lesions or scarring. If scarring was found and the individual had not mentioned herpes zoster in the medical history, they were asked for confirmation and details of the episode. Participants were then seen routinely every 3 months and at other times when they were ill. Doctors seeing the participants at routine appointments recorded the detailed medical history and physical examination on a questionnaire. Participants were asked if they had had herpes zoster since the last visit and the presence and the distribution of any scarring was recorded at each visit. The diagnoses from visits made between the 3 monthly appointments were also entered on the questionnaire. Clinical case notes were retained at the clinic but the questionnaires were double entered and validated at the main office 120 km away. The staff at the clinic were unaware of and, for reasons of confidentiality, unable to find out the HIV serostatus of any of the members of the cohort. However, all participants were encouraged to attend independent HIV testing and counselling services provided by the programme in the study villages.
Participants seen with acute herpes zoster received gentian violet paint and, because of the high rate of secondary bacterial infection found in this area, most received prophylactic antibiotics.
Since 1993, CD4 cell counts were measured every 6 months in HIV-positive and every year in HIV-negative participants using FACSCOUNT (Becton Dickinson, San José, California, USA).
Rates of herpes zoster in the cohort
The average rates of herpes zoster per person-year of observation (pyo)) according to different HIV subgroups were calculated using time from first to last visit. Participants who seroconverted in the cohort contributed observation time to the HIV-negative group until their first HIV-positive visit, after which they contributed to the HIV incident group.
Incidence of the first episode of herpes zoster since time from seroconversion
The incidence of the first episode of herpes zoster since time from seroconversion was investigated among incident cases using the Aalen–Nelson estimator. This estimator is analogous to the Kaplan–Meier survival estimator focusing on cumulative incidence. A Cox proportional hazards model was fitted using the incident cases in order to study the effect of CD4 cell count (included as a time-dependent covariate), age at enrolment and sex, on the time from seroconversion to the first episode of herpes zoster. Only incident cases were included in this analysis to avoid any bias owing to disease duration and the analysis only included time up to the participants’ last visit.
The effect of herpes zoster on progression to death
Finally, all the HIV-positive participants were investigated using a Cox model to assess the effect of having had a herpes zoster episode on progression to death; the model was adjusted for CD4 cell count (included as a time-dependent covariate) and other characteristics. Herpes zoster was also included as a time-dependent covariate (i.e., equal to 0 before the first episode and equal to 1 after the first episode). This allowed assessment of the effect of a past episode of herpes zoster on time to death. As this took into account all HIV-positive participants since their enrolment (prevalent cases) or since seroconversion (incident cases), this model was stratified according to these two different subgroups.
A CD4 cell count was performed on average every 6 months and so CD4 cell counts were not available for some patient visits. To be able to incorporate these visits in the analyses adjusted for CD4 cell counts, interpolation of the CD4 cell count was used. For remaining visits with missing CD4 cell counts the last CD4 cell count within 6 months of the patient's follow-up date was used. The CD4 cell counts were coded in three categories, < 200, 200–499 and > 500 × 106 cells/l, because the CD4 cell count is not necessarily log linearly related to the risk of death or herpes zoster.
All analyses were performed using STATA 6.0. statistical package (Stata Corporation, College Station, Texas, USA).
By the end of June 1999, 107 prevalent cases (infected with HIV prior to 1990), 144 incident cases, with seroconversion dates between mid-1990 and December 1998, and 231 negative participants had been enrolled. One prevalent and nine negative participants were not seen again after enrolment and so they were not included in the analyses. The age and gender of the participants with follow-up from first to last visit in the cohort are shown in Table 1.
Eight individuals had herpes zoster prior to their enrolment into the cohort; four had acute herpes zoster on enrolment (including one who was suffering from a second attack). Overall, 47 individuals were seen with 57 acute episodes of herpes zoster (Table 2). One of the HIV-negative controls seroconverted 5 months after developing herpes zoster. She had a negative HIV test [using two competitive enzyme-linked immunosorbant assays: Recombigen HIV-1/HIV-2 EIA (Cambridge Diagnostics Ireland Ltd, Ireland) and Wellcozyme HIV Recombinant EIA [VK57] (Abbot Murex Biotech Ltd, Dartford, UK)] and a CD4 cell count of 1505 × 106 cells/l at the time of the herpes episode and another unequivocal negative HIV test 15 weeks later. A repeat HIV test was positive (confirmed by Western blot) 16 weeks after the last negative test. Therefore, although there is the possibility that herpes zoster could have been a seroconversion event, the high CD4 cell count and repeat HIV-negative test 4 months later make this unlikely.
The median age [interquartile range (IQR)] at first herpes zoster episode in the prevalent, incident and negative groups was 38.1 (26.5–47.5), 29.9 (23.3–41.5) and 34.9 (32.7–55.4) years, respectively. There was no statistical difference in the ages (Kruskal–Wallis, P = 0.15) or gender distribution (Fisher's exact P = 0.9) between the HIV groups.
Fourteen prevalent and 19 incident patients with HIV infection and the five HIV-negative controls had a CD4 cells count measurement within 6 months of their first episode of herpes zoster. The median (IQR) CD4 cell count for individuals with herpes zoster in each HIV group was 320 (176–440); 489 (320–718) and 898 (452–1179) × 106 cells/l, respectively. The median CD4 cell count in the HIV-infected participants with herpes zoster was 437 × 106 cells/l, with a wide range of 47–1146 × 106 cells/l. Only five participants, four prevalent cases and one incident case, had a CD4 cell count of < 200 × 106 cells/l.
Eight (17.0%) out of the 47 participants with a first episode of herpes zoster had a second episode while in the cohort, three were prevalent and five were incident cases of HIV infections. The median (IQR) time between the first and second episodes was 29.3 (28.4–32.2) months. One incident patient had four episodes, with repeat episodes at 28.2, 37.2 and 49.7 months after the first. Four participants had recurrences affecting the same dermatomes, and three had a different dermatomal distribution from their first episode. The person with four episodes had the first two episodes affecting the same dermatomes, but the subsequent two attacks affected different dermatomes.
Post-herpetic neuralgia was reported by ten (all HIV positive) of the 47 participants, although in four this was only mentioned at the routine appointment following the episode and they did not attend with this complaint between the routine appointments. Another two participants attended the clinic just once as interim visits within 2 months of the episode. All of these received simple analgesics. Four patients had multiple clinic visits for many months following their episode of zoster. They received simple analgesics, codeine and pentazocine depending on the severity of their symptoms, and two received trials of amitriptyline.
The distribution of the herpes zoster lesions
The number of participants with lesions affecting particular dermatomes were two with cranial (both trigeminal), eight with cervical (one HIV negative), two with cervicothoracic (one HIV negative), 23 with thoracic (four HIV negative), six with thoracolumbar and six with lumbosacral. One person did not have this information recorded and the one HIV-negative control who seroconverted 5 months after developing herpes zoster had lesions simultaneously affecting thoracic dermatomes on the left side and cervical dermatomes on the right.
Rates of herpes zoster in the cohort
The mean rate [95% confidence interval (CI)] of herpes zoster from first to last visit was 53.6/1000 pyo (39.6–72.4) in the HIV-positive participants and 4.4/1000 pyo (1.8–10.6) in the HIV-negative controls. The rate of herpes zoster was higher in the prevalent than in the incident HIV-infected groups (63.1 and 45.3/1000 pyo, respectively); although this was not statistically significant (Mantel–Haenszel score test, P = 0.28), it might reflect the limited follow-up times in the groups. Cumulative probability of developing herpes zoster during follow-up did not vary between the prevalent and incident groups (log rank test, P = 0.33).
Incidence of herpes zoster after seroconversion
The median (IQR) length of follow-up from seroconversion to last visit for the 144 participants with incident HIV infection was 5.3 (2.8–7.2) years and, 22 had a first episode of herpes zoster a median (IQR) of 3.6 (1.1–5.0) years after seroconversion. The cumulative incidence of the first episode of herpes zoster since seroconversion is shown in Fig. 1. The cumulative rate shown seems roughly linear, implying that the rate of herpes zoster is constant from the time of seroconversion. The cumulative incidence of developing a first episode of herpes zoster was 7.6% at 2 years, 12.6% at 4 years,and 24.0% at 6 years after seroconversion and the overall incidence rate was 35.6/1000 pyo (95% CI 23.4–54.0).
In order to estimate the effect of several characteristics on time from seroconversion to herpes zoster, 1665 visits among these 144 individuals were taken into account. There was no evidence of a significant effect of age, gender and CD4 cell count on the incidence of the first episode of herpes zoster, as shown in Table 3.
The effect of herpes zoster on progression to death
There were 92 deaths in the 250 HIV-positive participants from the beginning of follow-up to end of June 1999. Results of the stratified univariate survival analysis are presented in Table 4. An episode of herpes zoster was a risk factor for death if taken alone (P = 0.04). CD4 cell count and age were the strongest risk factors for death, and after adjusting for the effect of these, the association between a first episode of herpes zoster and a faster progression to death was no longer significant (P = 0.45).
The rates and disease manifestations of herpes zoster have been described in a clinical cohort in rural Uganda, comparing the findings in HIV-positive and HIV-negative individuals (although it was an unusual event in the controls) and examining the incidence of herpes zoster after seroconversion and whether it was a prognostic indicator of a more rapid death. This was an unbiased report of episodes in the population since participants were reviewed every 3 months and attended the clinic for treatment of illnesses occurring between routine appointments. Consequently, it is unlikely that any episodes of herpes zoster were missed during follow-up. Many reported studies of herpes zoster used attendances at specialist or referral clinics and, therefore, missed those people who did not seek medical help. This is especially important in sub-Saharan Africa where people often cannot afford treatment costs or prefer to use traditional medicine. There have been few studies reporting herpes zoster in HIV-infected people in Africa and, as far as we know, none of these has been population based. However, hospital or referral clinic studies have found that HIV-positive patients with herpes zoster had a longer duration of lesions, more generalized lymphadenopathy, severe pain, superadded bacterial infection and more than one affected dermatome compared with HIV-negative patients. These studies also reported that the distribution of lesions was similar in HIV-infected and uninfected patients, except for cranial nerve involvement, which was only seen in HIV-positive patients [5,8].
In our study, the rate of new herpes zoster episodes in the HIV-positive participants was 54/1000 pyo; although this is higher than that reported in some studies [2,11,12], it is within the range 25–55/1000 pyo reported overall [3,12,13]. Rates reported in HIV-infected groups depend on the cohort or centre collecting the data and the frequency and completeness of follow-up. Rates have been found to differ in homosexual men and haemophiliacs and to decrease over time . The differences between our rates and other studies could also reflect our intensive follow-up of participants and the 3-monthly review at which participants were not only asked if they had had herpes zoster but were also examined for scarring. The rate in the HIV-negative controls was 4/1000 pyo and this is similar to 2–3.3/1000 pyo reported in other cohorts [2,3]. Our rates in the HIV-infected participants were over 10 times higher, agreeing with the finding that in younger people herpes zoster is an indication of HIV infection [5–7]. Our recurrence rate of herpes zoster of 17% during follow-up was higher than those reported by some studies [4,13] but lower than those reported by other groups (22–27%) [2,3,5,11]. Obviously, the number of people who developed a recurrence depends on the period of follow-up. The median period between first episode and a recurrence was longer than that previously reported [11,12].
Our cumulative incidence of 24% at 6 years after seroconversion is higher than 7% reported by Alliegro et al.  but is very similar to 20–23% reported from homosexual cohorts [2,3,13]. The rate did not increase with time from seroconversion. Most groups agree that herpes zoster is not associated with duration of HIV infection and the cumulative incidence increases linearly with duration of follow-up, showing that it is neither an early nor late manifestation but occurs at a relatively constant rate following infection [2,3]. This fits with our finding that herpes zoster was not a marker of deteriorating immune function as indicated by CD4 cell counts. CD4 cell count was not a risk factor for an episode of herpes zoster and the range of CD4 cell counts at which participants presented with their first episode of herpes zoster was wide. There has been disagreement in the literature regarding whether herpes zoster increases with decreasing CD4 cell counts. Some groups reported finding no relationship [2,13] while others have reported that the CD4 cell counts were lower in those with herpes zoster  and that rates of herpes zoster episodes increased in patients with lower CD4 cell counts [3,12]. Our mean CD4 cell count in the HIV-positive group was higher, and a greater proportion had a CD4 cell count of > 500 × 106 cells/l than at a referral clinic in Kenya where herpes zoster was often recognized as an initial HIV-related illness despite moderate to severe depression of CD4 cell counts .
We found that herpes zoster was not a clinical marker of a more rapid progression to death after adjustment for age and CD4 cell count. Again, there has been conflicting evidence in previously published papers regarding this issue, with some groups reporting no association after adjustment for CD4 cell counts [2–4,13–15] and others finding one [5,16,17]. The differences in these results may be explained by the influence of CD4 cell counts. In our paper, we take into account the relative change of CD4 cell counts during the follow-up of an individual.
Although we have achieved reasonable follow-up for the HIV-positive group overall, the incident cases from seroconversion and the number of events of herpes zoster and death, there may be a power limitation when intergroup analyses are performed. This would particularly affect the estimation of the time from seroconversion to first episode of herpes zoster, because there are relatively few events. However, we can assume that there is minimal power limitation using time to death as the endpoint, because the number of events is higher. In order to avoid bias in analyses using adjusted CD4 lymphocyte counts, interpolation (when the missing CD4 cell count is between two available CD4 cell counts) and not extrapolation (when a missing CD4 cell count precedes or follows any available CD4 cell counts) was used to incorporate visits with missing CD4 data. This also minimized the bias of possible fluctuations of CD4 cell counts. The extent of the interpolation was about 53% of the follow-up visits, because a CD4 cell count was performed on average every 6 months. The grouping of CD4 cell counts into three categories should further reduce the bias of interpolation.
In conclusion, we found that rates of herpes zoster in HIV-infected persons, the cumulative proportion developing initial episodes after seroconversion, the recurrence rates and the clinical presentation of herpes zoster in a rural population in Uganda were similar to those reported from industrialized countries. The development of herpes zoster was unrelated to CD4 cell count or period from seroconversion. Herpes zoster was an indicator of HIV-1 infection in this population but not an indicator of more rapid progression after adjusting for CD4 cell count and age.
We wish to thank all the clinic staff and the participants in the study.
1. Cull RE, Will RG. Diseases of the nervous system.
In: Davidson's Principles and Practice of Medicine. Edited by Edwards CRW, Bouchier IAD, Haslett C, Chilvers ER. Edinburgh: Churchill Livingstone; 1995: 1091 1091.
2. Buchbinder SP, Katz MH, Hessol NA. et al. Herpes zoster and human immunodeficiency virus.
J Infect Dis 1992, 166: 1153 –1156.
3. Veenstra J, Krol A, van Praag RME. et al. Herpes zoster, immunological deterioration and disease progression in HIV-1 infection.
AIDS 1995, 9: 1153 –1158.
4. Rogues AM, Dupon M, Ladner J, and Groupe d'Epidémiologe Clinique du SIDA en Aquitaine. Herpes zoster and human immunodeficiency virus infection: a cohort study of 101 coinfected patients.
J Infect Dis 1993, 168: 245. 245.
5. Colebunders R, Mann JM, Francis H. et al. Herpes zoster in African patients: a clinical predictor of human immunodeficiency virus infection.
J Infect Dis 1988, 157: 314 –318.
6. Lindan CP, Allen S, Serufilira A. et al. Predictors of mortality among HIV-infected women in Kigali, Rwanda.
Ann Intern Med 1992, 116: 320 –328.
7. Hira SK, Ngandu N, Wadhawan D. et al. Clinical and epidemiological features of HIV infection at a referral clinic in Zambia.
J Acquir Immune Defic Syndr 1990, 3: 87 –91.
8. Tyndall MW, Nasio J, Agoki E. et al. Herpes zoster as the initial presentation of Human immunodeficiency virus type-1 infection in Kenya.
Clin Infect Dis 1995, 21: 1035 –1037.
9. Mulder DW, Nunn AJ, Wagner U-H, Kamali A, Kengeya-Kayondo JF. HIV-1 incidence and HIV-1-associated mortality in a rural Ugandan population cohort.
AIDS 1994, 8: 87 –92.
10. Morgan D, Malamba SS, Maude GH. et al. An HIV-1 natural history cohort and survival times in rural Uganda.
AIDS 1997, 11: 633 –640.
11. Glesby MJ, Moore RD, Chaisson RE. Clinical spectrum of herpes zoster in adults infected with human immunodeficiency virus.
Clin Infect Dis 1995, 21: 370 –375.
12. Engels EA, Rosenberg PS, Biggar RJ, for the District of Columbia Gay Cohort Study and the Multicenter Hemophilia Cohort Study. Zoster incidence in human immunodeficiency virus-infected hemophiliacs and homosexual men, 1984–1997.
J Infect Dis 1999, 180: 1784 –1789.
13. McNulty A, Li Y, Radtke U. et al. Herpes zoster and the stage and prognosis of HIV-1 infection.
Genitourin Med 1997, 73: 467 –470.
14. Alliegro MB, Dorrucci M, Pezzotti P. et al., and the Italian HIV Seroconversion Study Group. Herpes zoster and progression to AIDS in a cohort of inividuals who seroconverted to human immunodeficiency virus.
Clin Infect Dis 1996, 23: 990 –995.
15. Glesby MJ, Moore RD, Chaisson RE, and the Zidovudine Epidemiology Study Group. Herpes zoster in patients with advanced human immunodeficiency virus treated with zidovudine.
J Infect Dis 1993, 168: 1264 –1268.
16. Melbye M, Grossman RJ, Goedert JJ, Eyster ME, Biggar RJ. Risk of AIDS after herpes zoster.
Lancet 1987, 1: 728 –731.
17. van Griensven GJP, de Vroome EMM, de Wolf F, Goudsmit J, Roos M, Coutinho RA. Risk factors for progression of human immunodeficiency virus (HIV) infection among seroconverted and seropositive homosexual men.
Am J Epidemiol 1990, 132: 203 –210.