Page 259 - SAHCS HIVMed Journal Vol 20 No 1 2019

Page 3 of 8 Original Research

prevent adhesions. Pain was controlled with Tramadol Results
hydrochloride 50 mg – 100 mg 6 hourly and paracetamol 1 g
6 hourly. Pethidine was used to alleviate the pain while Of the 36 patients in the study, 32 (88.9%) were women.
dressing the wounds. Cutaneous lesions were managed by Eleven had SJS, 13 had SJS/TEN overlap and 12 had TEN.
strict barrier protection nursing and meticulous wound care The number of days of hospitalisation increased exponentially
without debridement using nanocrystalline silver dressings along the spectrum of the disease (Table 2).
(Acticoat®). Antibiotics were not used prophylactically
unless there was a clinical indication. Expert opinion Sixteen (50.0%) women in the study were pregnant. The
was sought from the ophthalmologists on admission and mean (s.d.) CD4 cell count in the pregnant women was
3
instructions were carried out as per the ophthalmologist care 267.2 (60.6) cells/mm . A significant number of pregnant
plan. Their management plan entailed the use of a topical women (93.8%) developed SJS/TEN secondary to nevirapine,
steroid (Maxitrol®), lubricants and glass rodding to prevent while one was due to isoniazid prophylaxis. The pregnant
adhesions. Obstetricians and physicians were consulted women presented at a mean (s.d.) of 29.13 (3.76) weeks
when indicated. gestation. Five (31.25%) of the women delivered prematurely
as a result of foetal distress, and all five of these patients had
TEN. One of the deliveries was a stillbirth at 34 weeks’
Data analysis gestation. The other four infants were healthy and showed

Data were analysed using Stata 13.0 SE (StataCorp. 2013. no signs of any drug reaction.
Stata Statistical Software: Release 13. College Station, TX:
StataCorp LP). Descriptive statistics included means CD4 cell counts were available for all the patients but one
(standard deviations) for continuous variables and (Table 3). There was no association between the CD4 cell
frequencies (n [%]) for categorical variables. Relationships count and SCORTEN score (Spearman rho correlation test of
between continuous predictors and dichotomous outcomes 0.039). There was no significant difference in the mean CD4
were assessed using the standard t-test or non-parametric cell count by drug type with an ANOVA p-value of 0.594
Wilcoxon rank-sum test if the normality assumptions were (data not shown).
violated. Similarly, analysis of variance (ANOVA, or non-
parametric equivalent, namely the Kruskal–Wallis equality- The mean (s.d.) CD4 cell count and confidence interval
of-populations rank test) was employed to compare the noted in patients with complications were 236.8 (186.4–287.2)
3
means of continuous predictors across the three-drug cells/mm . There was no statistical difference across all
reaction groups. Correlation between continuous variables complications. Thus, the complications seen in the patients
was assessed using the Spearman rank correlation coefficient. with SJS/TEN were not influenced by the CD4 cell counts
Differences in frequencies of categorical explanatory and hence the level of immunosuppression in the sample.
variables by drug reaction category were assessed using the
2
Pearson chi-square (χ ) test or Fisher’s exact test if an expected The average SCORTEN scores for SJS, SJS-TEN overlap and
cell count contained fewer than five observations. A p-value of TEN were 1, 2 and 3, respectively. There was a significant
< 0.05 was deemed statistically significant. difference in the median SCORTEN score by drug reaction
type p < 0.001. Thus, the more severe the drug reaction, the
Ethical consideration higher the SCORTEN score observed.

Ethical approval was obtained from the University of Comorbidities included hypertension, tuberculosis and
KwaZulu-Natal Biomedical Research Ethics Committee epilepsy (Table 3). Ten (27.8%) of the patients in the
(reference number BE417/14). study reacted to the anti-tuberculosis therapy, two due to

TABLE 2: Demographics of 36 HIV-infected patients presenting with Stevens–Johnson syndrome–toxic epidermal necrolysis.
Variable Total (n = 36) SJS (n = 11) SJS/TEN (n = 13) TEN (n = 12)
n % s.d. Mean n % s.d. Mean n % s.d. Mean n % s.d. Mean
Age (years)
< 40 years 30 83.3 - - 10 27.8 - - 10 27.8 - - 10 27.8 - -
> 40 years 6 16.7 - - 1 2.8 - - 3 8.3 - - 2 5.6 - -
Mean (s.d.) age
Male - - 7.6 34.3 - - - 34.3 - - - - - - - -
Female - - 10.4 32.4 - - - 27.1 - - - 34.8 - - - 33
Sex
Male 4 11.1 - - 4 11.1 - - 0 - - - 0 - - -
Female 32 88.9 - - 7 19.4 - - 13 36.1 - - 12 33.3 - -
Length of hospital stay (Mean s.d.) no. days - - 11.2 15.1 - - 2.4 5.8 - - 6.9 13.0 - - 11.3 25.8
Pregnant patients (n = 32)
Y 16 50 - - 5 31.3 - - 5 31.3 - - 6 37.5 - -
N 16 50 - - 2 12.5 - - 8 50 - - 6 37.5 - -
SJS, Stevens–Johnson syndrome; TEN, Toxic epidermal necrolysis; SJS, TEN overlap; Y, yes; N, no; Regimen 1: NVP, nevirapine; 3TC, lamivudine; D4T, stavudine; TMP/SMX, trimethoprim/
sulfamethoxazole; TB, tuberculosis; N/A, not applicable.

http://www.sajhivmed.org.za 252 Open Access