Dear editor,

We read with great excitement the article by Perel et al. where they presented a model for predicting risk of early death in trauma patients with or at risk for significant bleeding [1]. Especially commendable was the authors attempt to include potential end users in the development process, and also coming up with a point-of-care tool including an online calculator. This approach is unfortunately far too rare. Furthermore, the authors proposed three different models, one for each country-level income category.

While acknowledging the differences in trauma outcomes between high-, middle-, and low-income countries is definitely a step in the right direction, the World-bank categories unfortunately do not tell the whole truth. Many countries, high-, middle-, and low-income countries show a wide internal diversity in terms of income distribution, hence it is likely that also trauma outcomes differ widely within the country depending on the patient population catered to in combination with trauma center setup. This should be particularly true for low- and middle-income countries with huge populations and big gaps in income distribution such as India.

The article by Perel et al. [1] was of particular interest to us because Lokmanya Tilak Municipal General Hospital (LTMGH) was one of the collaborating partners in the CRASH 2 trial. This previous collaboration made us keen to test the proposed model on a recently collated trauma dataset. Lokmanya Tilak Municipal General Hospital is a public hospital providing subsidized health care. It is located in the center of megapolis Mumbai, and caters to a population of which 70% lives in some of Asia’s biggest slums. Hence, our hypothesis was that even though India is a middle-income country, the trauma outcomes at LTMGH might be better captured using the low-income model. Our dataset included 1119 blunt and/or penetrating trauma patients, of which 202 were between 15 and 81 years old and had significant bleeding according to the CRASH 2 definition (on admission systolic blood pressure<90 mmHg or heart rate>110 or both) [2]. The results of applying the CRASH 2 model to our dataset are presented in table 1.

Our findings show that of the three models proposed by Perel et al. [1], the high-income model performed the worst, while our hypothesis that the low-income model would correspond with the reality held true. Hence, we would like to stress the importance of adopting a systems view and admitting that while national level classifications might be useful on a broad scale, point-of-care usage of models such as this must take into account intra-country variations.

References:
1. Perel, P., et al., Predicting early death in patients with traumatic bleeding: development and validation of prognostic model. BMJ, 2012. 345: p. e5166.
2. Shakur, H., et al., Effects of tranexamic acid on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. Lancet, 2010. 376(9734): p. 23-32.

Affiliations:
MG: Division of Global Health (IHCAR)
Karolinska Institutet
17177 Stockholm, Sweden

VK: Department of Surgery.
L.T.M.M C. & L.T.M.M. Hospital
Sion, Mumbai – 400022. India.

MK: Department of Surgery
Seth G.S. Medical College & KEM Hospital
Parel, Mumbai – 400016. India.

SD: Department of Surgery.
L.T.M.M C. & L.T.M.M. Hospital
Sion, Mumbai – 400022. India.

NR: Dept. of Surgery
BARC Hospital
Mumbai, India 400094

Competing interests: None declared