HELLP syndrome: utility of specific classifications as prognostic tools

Introduction

HELLP syndrome is a specific complication of pregnancy characterized by hemolysis, elevated liver enzymes and low platelet count. Maternal mortality was reported to be as high as 24%. Two classifications of the HELLP syndrome are widely used (Tennessee [1] and Mississippi [2]). The aim of this study is to determine mortality of HELLP syndrome as defined by each classification and try assessing the most relevant.

Patients and methods

Prospective data collection as part of the APRiMO study (Assessment of Prognosis and Risk of Mortality in Obstetrics). Included were all obstetric patients transferred from a referral center for high-risk pregnancies in our independent multidisciplinary ICU. The study period was January 1996–September 2004. Demographic data, obstetric history, morbid events, length of stay (LOS), severity of illness scoring systems and organ dysfunction scores at day 1 of admission were collected. Exclusion criteria were LOS < 4 hours. The main outcome of interest was survival status at ICU discharge. Two groups were compared: patients with HELLP syndrome as defined alternatively by the two classifications (Group I), and patients without hepatic dysfunction (Group II). Results are expressed as the mean ± standard deviation. P < 0.05 was considered significant. Discrimination of the classifications was assessed by the area under the receiver operator characteristic curve (AuROC). Calibration was assessed by the Hosmer–Lemeshow (HL) goodness-of-fit test. Data were computed on SPSS 11.5, Win-XP compatible.

Results and discussion

Differences between Group I and Group II were statistically significant concerning obstetric hemorrhagic complication (P < 0.001), incidence of acute renal failure (P = 0.01), mortality (P = 0.001), LOS (6.5 ± 7 days vs 4.4 ± 4 days, P = 0.001), SAPS-Obst score (24.5 ± 8 vs 16.8 ± 7, P < 0.001). The Mississippi classification discriminated well, but calibrated badly. In contrary, the Tennessee classification was a poor discriminator but calibrated very well. See Table 1.

Table 1 Number of patients, discrimination and calibration statistic tests for each classification

Conclusion

Both models classified patients according to different criteria but were correlated with mortality. None of the classifications discriminated and calibrated well at the same time. The two models seem to be complimentary. Development of an aggregate classification could refine the models.