3.1. Survey of distribution
From a total of 771 patients scrutinised (age average: 32.54 ± 23.2 years) including 332 women and 439 men, 58 patients were admitted in the emergency room of the Claudio Benatí Hospital with evident symptoms of HAPE after sojourning at 3,600-4,000 m.a.s.l. in Zumbahua. The age average of the HAPE group (13 women and 45 men) was 34.57 ± 27 years. Within the data frame, a total of 448 people were self-perceived as indigenous and 323 as mestizos. All data sampled at 95% confidence level has a margin of error of 5.3%.
3.2. Analysis of HAPE risk factors
Demographic factors like sex and age, altitude of permanent residence (> 3,000 m.a.s.l.), and the MCHC blood parameter was significantly associated with HAPE susceptibility (see Fig. 1, Fig. 2 and Table 1).
Figure 1 remarkably represents the difference due to sex and altitude. MCHM in males living over 3000 m.a.s.l. is higher by 7% compared to the grand mean of the data frame (N = 771). In contrast, in females, MCHC differences across altitudes did not reach statistical significance. Also according to the GLM model, the rest of the factors analysed (ethnicity, RBCs, HCT, HGB; MCV, MCH; blood pressure, heart rhythm, breathing rate, and blood O2 saturation), turned out irrelevant to HAPE susceptibility (data not shown).
In Eq. 2. (shown below), the intercept for the model was settled with SEXFemale, MCHC (continuous), ALTITUDE< 3000 and AGE25 − 64 as starting points; odds linked to this combination of factors was ≈ 1e− 19. These conditions pointed to the lowest odds for the cohort under analysis. Conversely, the following subset of conditions: being male, over 65 years old, with MCHC = 37 and residing > 3,000 m.a.s.l. in the Andean highlands, led to odds of ≈ 1.89, that is an increase of risk compared with the base condition of the model. The above was graphically represented in Fig. 2 as a function of MCHC.
Equation 2. Logit model, representing the logarithmic odds associated with HAPE condition.
$$ln \left(\frac{p\left(HAPE\right)}{1-p\left(HAPE\right)}\right) =-43.69+1.78*{SEX}_{male}+0.66*MCHC+15.93*{ALTITUDE }_{>3000}+$$
$$2.21*{AGE}_{0-13}+ 1.75*{AGE}_{14-25}+ 21.18*{AGE}_{>65}$$
1
Table 1
Analysis of deviance for the GLM model fitted.
Variable | df | Deviance | Residual df | Residual deviance | p-value (> chisqrt) |
Null | | | 231 | 94.452 | |
Sex | 1 | 7.5273 | 230 | 86.925 | 0.0061 |
MCHC | 1 | 10.0405 | 229 | 76.884 | 0.0015 |
Altitude | 1 | 2.8498 | 228 | 74.035 | 0.0913 |
Age | 3 | 6.4774 | 225 | 67.557 | 0.0905 |
The objective of the GLM analysis was to find an equation that best predicted the odds of HAPE occurrence and its association with the variables observed in the Eq. 2. Such an equation may provide information about their attributable fraction or weight and the deviance linked to any factor of interest. The analysis was only conducted on those significant variables that had an annotated p-value lower than 0.05 in the ANOVA of the models (data not shown). The variables statistically significant were Sex, Altitude, MCHC and Age for the fit model (Table 1). The residence at low altitude was inversely related to the likelihood of having HAPE. Nevertheless, given the influence of the other demographic variables, the multiple logistic regression analyses reflected the dependence of HAPE susceptibility on age. Female sex as well as age were negatively related to the likelihood of having HAPE, whereas the indigenous ethnicity was positively related, yet no statistically significant. The relative influences on HAPE risk were as follows: MCHC > Sex > Altitude > Age (≥65) ≃ Age (0−13) > Age (14−25) ≃ Age (26−64).
The analysis of deviance indicates the relevance of Sex as a risk factor for HAPE in highlands habitants and the MCHC as a consistent haematological biomarker for predictions of this condition. In Fig. 2, the black lines were added based on an arbitrary threshold odds ratio of 1.05 and considering values over this threshold as significant for MCHC increment. Complementary, a MCHC value of 34 g/dL was adopted as the reference to analyse the behaviour and correlation for the subjects under study. The latter suggests that a linear increment of MCHC may result in an exponential increase in the risk of HAPE.