Sepsis is defined as life-threatening organ dysfunction arising from a dysregulated host response to infection, and septic shock as the subset in which circulatory, cellular and metabolic abnormalities are profound enough to substantially increase mortality [1]. It remains among the leading causes of death in non-coronary intensive care units worldwide, and despite four decades of advances in antimicrobial therapy and organ support, both incidence and absolute mortality have continued to rise [2]. Reported case fatality varies widely with setting and definition, approaching 80% for septic shock in some North American series and 45.7% in European data, with reporting from developing countries likely to understate the true burden [2,3].

Metabolic acidosis is one of the most consistent biochemical accompaniments of the septic shock state. It is characterised by a primary fall in serum bicarbonate, a compensatory reduction in arterial pCO₂ and a consequent reduction in blood pH, and among critically ill patients it commonly forms part of a mixed disturbance rather than occurring in isolation [4]. The pathophysiological drivers in sepsis are several and overlapping: microcirculatory failure with impaired oxygen extraction, anaerobic glycolysis with lactate accumulation, acute kidney injury with reduced net acid excretion, hepatic dysfunction with impaired lactate clearance, and iatrogenic hyperchloraemia arising from large-volume chloride-rich crystalloid resuscitation [5,6]. Gunnerson has emphasised that the clinical meaning of an acid–base abnormality in the intensive care unit depends heavily on which of these mechanisms predominates [7].

The prognostic value of the individual ABG components has attracted sustained interest because they are inexpensive, rapidly obtained and available at the bedside long before culture data or biomarker panels return. Smith and colleagues, in a landmark study of patients admitted to intensive care, established that base excess and lactate function as prognostic indicators independent of admission diagnosis [8]. Kaplan and Kellum subsequently demonstrated that initial pH, base deficit, lactate and anion gap each predicted outcome after major vascular injury [9]. Dunham et al. had earlier framed metabolic acidaemia as a quantitative surrogate for accumulated oxygen debt and thus for the severity of the ischaemic insult [10]. In a large multicentre prospective analysis, Jung and colleagues found severe metabolic acidaemia in 6% of intensive care admissions, with mortality of 57% in that group, and showed that both the depth of the acidaemia and the time taken for it to correct were linked to survival [11].

Base excess deserves particular attention among these variables. It quantifies the metabolic component of an acid–base disturbance independently of respiratory compensation, and Davis et al. showed that admission base deficit predicts transfusion requirement and complication risk in trauma [12]. Whether it retains comparable discriminatory power in septic shock specifically, and how it compares with pH, pCO₂ and bicarbonate in the same cohort, is less well established in the Indian literature. Kiran et al. addressed severe metabolic acidosis and outcome in a South Indian intensive care population and reported substantial mortality differences [13], but corroborating data restricted to septic shock remain limited.

The present study was therefore undertaken to characterise the admission acid–base profile of critically ill patients with septic shock at a tertiary care teaching hospital, and to compare pH, pCO₂, bicarbonate and base excess between survivors and non-survivors in order to establish which of these immediately available parameters best identifies patients at highest risk of death.

Study design and setting. This was a prospective observational study conducted in the Medical Intensive Care Unit (MICU) of K.R. Hospital, attached to Mysore Medical College and Research Institute, Mysuru, Karnataka. The unit functions as a referral intensive care facility for a large catchment population in southern Karnataka and receives both direct emergency admissions and internal transfers from medical wards. Study duration. Data were collected prospectively over twelve consecutive months, from January 2018 to December 2018. Sample size. The required sample was estimated using the formula 4pq/d², where p represented the prevalence of community-acquired pneumonia recorded at the institution over the preceding year, q = 1 − p, and d = 1.5. This computation yielded 97, which was rounded to a final sample size of 100. Participants. Consecutive patients admitted to the MICU who satisfied the criteria for sepsis and septic shock as defined by the Surviving Sepsis Campaign International Guidelines (2018 update) were screened. Inclusion required an age above 18 years and written informed consent from the patient or an accompanying relative. Patients were excluded if they were pregnant, had a diagnosed malignancy, were HIV-positive and receiving treatment, or had established chronic liver disease. These exclusions were applied because each condition independently perturbs acid–base handling or lactate clearance and would have confounded interpretation of the admission blood gas. Data collection. For every enrolled patient, age, sex, presenting symptoms and signs, working diagnosis, anticipated source of infection, treatment administered including intravenous fluid therapy, duration of hospital stay and final outcome were recorded on a structured proforma. A single arterial blood gas sample was drawn within 24 hours of MICU admission and analysed for pH, pCO₂, bicarbonate, lactate, base excess and anion gap. The Sequential Organ Failure Assessment (SOFA) score was computed for each patient from the corresponding clinical and laboratory variables. Supporting investigations comprised complete haemogram, liver function tests, renal function tests, serum electrolytes, random blood sugar and electrocardiography. Cultures of blood, urine, sputum and pus were obtained as clinically indicated. Definitions and stratification. For analysis, blood pH was categorised as below 7.20, 7.21–7.35, and above 7.35. Partial pressure of carbon dioxide was grouped as below 35 mmHg, 35–45 mmHg, and above 45 mmHg. Bicarbonate was stratified as below 22 mEq/L, 22–28 mEq/L, and above 28 mEq/L. Base excess was categorised as positive, down to −5, −6 to −10, −11 to −15, and −16 to −20 mEq/L. Outcome measure. The primary endpoint was in-hospital outcome, dichotomised as survival to discharge by the treating physician or in-hospital death. Patients were followed from enrolment until one of these endpoints was reached. Statistical analysis. Data were entered and analysed using the Statistical Package for the Social Sciences version 20. Continuous variables are expressed as mean ± standard deviation and categorical variables as frequencies with percentages. Comparisons between the survivor and non-survivor groups employed the non-parametric t-test and analysis of variance as appropriate. A p value below 0.05 was taken to indicate statistical significance. Ethical considerations. Ethical clearance was obtained from the Institutional Ethics Committee of Mysore Medical College and Research Institute before commencement. Written informed consent was obtained from every participant or their accompanying attendant, participation was voluntary with an unrestricted right of withdrawal, and confidentiality of records was maintained throughout.

Table 1. Demographic profile and source of infection by outcome (n = 100)

Mean age 51.7 ± 17.2 years overall; 49.7 ± 2.5 years in males and 53.6 ± 2.5 years in females. Percentages are column percentages within outcome group.

Overall in-hospital mortality was 53%. Age was distributed almost evenly across the three bands and showed no gradient with outcome, each stratum contributing between 32% and 36% of the cohort with survivor and non-survivor proportions within a few percentage points of one another. Sex, by contrast, showed a marked asymmetry: although the cohort was exactly balanced at 50 males and 50 females, 31 of 50 males died against 22 of 50 females, giving a male mortality of 62% versus 44% in females. Lower respiratory tract infection was both the commonest source overall (42%) and disproportionately represented among deaths, accounting for 49.1% of non-survivors. The contrast between urogenital and soft tissue sources is striking in the opposite directions — urogenital infection accounted for 31.9% of survivors but only 9.4% of deaths, while soft tissue infection accounted for 15.1% of deaths but a single survivor.

Table 2. Admission blood pH by outcome

The separation between groups is among the sharpest in the dataset. Of the 48 patients presenting with a pH below 7.20, 43 died — a mortality of 89.6% within that stratum. Conversely, of the 23 patients with a pH above 7.35, only 3 died. The mean difference of 0.18 pH units may appear numerically modest, but because pH is a logarithmic scale this corresponds to approximately a 50% higher hydrogen ion concentration in non-survivors. The distribution also illustrates that acidaemia was not merely a marker of illness but stratified the cohort almost cleanly: 89% of non-survivors presented with a pH at or below 7.35, against 57% of survivors.

Table 3. Admission partial pressure of carbon dioxide by outcome

Non-survivor count in the < 35 mmHg stratum reconstructed from the reported proportion of 98.1%.

Hypocapnia dominated the cohort, affecting 85% of all patients, and was virtually universal among those who died — only one non-survivor had a pCO₂ within the normal range and none was hypercapnic. The mean pCO₂ of 18.8 mmHg in non-survivors represents profound hyperventilation. This finding must be read as a compensatory rather than a primary phenomenon: a low pCO₂ in the presence of a low pH and low bicarbonate indicates respiratory compensation for a primary metabolic acidosis. The degree of hypocapnia therefore functions as an indirect index of the severity of the underlying metabolic derangement, and its magnitude in the non-survivor group signals that compensatory mechanisms were being driven to their physiological limit.

Table 4. Admission serum bicarbonate by outcome

Non-survivor count in the < 22 mEq/L stratum reconstructed so that group totals reconcile to 53.

Bicarbonate depletion was the rule, present in 86% of the cohort, but its depth differentiated outcome. The mean value of 12.9 mEq/L among non-survivors represents loss of nearly half the normal buffer reserve, whereas the survivor mean of 19.3 mEq/L, though subnormal, retained substantially more capacity. Notably, a bicarbonate below 22 mEq/L was too common to be discriminatory on its own — three-quarters of survivors also met that threshold. It is the magnitude of the depletion, captured by the group means, rather than the presence or absence of the abnormality, that carries the prognostic signal.

Table 5. Admission base excess by outcome

Base excess produced the cleanest dichotomy of any parameter examined. Forty-eight of 53 non-survivors (90.6%) had a base deficit exceeding 11 mEq/L, against only 8 of 47 survivors (17.0%). Read in the other direction, of the 44 patients with a base deficit of 10 mEq/L or less, 39 survived; of the 56 patients with a deficit beyond that threshold, 48 died. A single cut-point at −11 mEq/L therefore correctly classified 87 of 100 patients in this cohort. The mean difference of more than 10 mEq/L between groups is the widest absolute separation among the acid–base variables, which is physiologically coherent given that base excess quantifies the metabolic disturbance while stripping out the respiratory compensation that confounds pH and bicarbonate interpretation.

Table 6. Comparison of acid–base parameters with published series

The direction and approximate magnitude of every comparison are reproduced across all four series, which is reassuring for external validity. The present cohort was younger than the two Indian comparators and considerably more balanced by sex, the 1:1 ratio contrasting with the male predominance of 2:1 and 3.5:1 reported elsewhere. Absolute pH values in the present study sit slightly higher in both outcome groups than in the comparator series, but the survivor-to-non-survivor gradient of roughly 0.2 pH units is common to all. Base excess in non-survivors is remarkably consistent at −14.0 to −17.0 mEq/L across three independent cohorts, reinforcing the impression that this parameter behaves reproducibly as a severity index.

The central observation of this study is that the depth of metabolic acidosis present at the point of intensive care admission stratifies septic shock patients into groups with radically different survival. Mortality in this cohort was 53% overall, but reached 89.6% among those presenting with a pH below 7.20 and 85.7% among those with a base deficit beyond 11 mEq/L. Each of the four acid–base parameters examined differed significantly between outcome groups at p < 0.001, and the separation was widest for base excess.

That base excess should outperform its companions is physiologically explicable. Both pH and bicarbonate are influenced by the respiratory response to the metabolic insult, and in a cohort where 85% of patients were hypocapnic, this compensation systematically compresses the apparent difference between groups. Base excess isolates the metabolic component, and its behaviour here mirrors that reported by Smith and colleagues, who found base excess and lactate to be prognostic indicators in a general intensive care population independently of admission category [8]. Kaplan and Kellum reached a comparable conclusion for initial pH, base deficit, lactate and anion gap in major vascular injury [9], and Davis et al. demonstrated the predictive utility of admission base deficit for complications and transfusion requirement in trauma [12]. Dunham and colleagues offered the unifying interpretation, framing metabolic acidaemia as a quantitative expression of accumulated oxygen debt and therefore of the magnitude of the ischaemic insult sustained [10].

The pCO₂ findings warrant careful reading. A mean of 18.8 mmHg among non-survivors represents extreme hyperventilation, and it would be a misinterpretation to regard low pCO₂ as favourable simply because it raises pH. In the setting of a low pH and a bicarbonate of 12.9 mEq/L, the hypocapnia is compensatory, and its extent indexes how far respiratory reserve has been recruited to defend systemic pH. A patient hyperventilating to this degree has little remaining capacity, and any subsequent respiratory fatigue or need for controlled ventilation risks abrupt decompensation. The finding that only one non-survivor had a normal pCO₂ and none was hypercapnic suggests that hypercapnia was largely absent as a presenting picture in this population.

The comparison with published series supports generalisability. Jung et al., in the largest prospective multicentre analysis of severe acidaemia in intensive care, reported 57% mortality in that subgroup, closely matching the 53% observed here, and demonstrated that both the severity of the acidaemia and the time required for its correction predicted survival [11]. Kiran et al., working in a comparable South Indian setting, reported almost identical base excess values in non-survivors [13]. Gunnerson has argued that the clinical significance of an acid–base abnormality depends on its mechanism [7], and Noritomi and colleagues, using longitudinal quantitative analysis in severe sepsis and septic shock, showed that the composition of the acidosis evolves over the course of resuscitation with lactate, unmeasured anions and hyperchloraemia contributing in shifting proportions [14]. Mecher et al. had earlier drawn attention to unaccounted anions during severe sepsis in humans [15].

Several limitations qualify these conclusions. The study was purely observational with no intervention. Critically, a single arterial blood gas obtained within 24 hours of admission formed the basis of analysis, with no serial or follow-up sampling; since the trajectory of acidosis correction is itself prognostic [11], this design cannot distinguish patients whose derangement resolved rapidly from those in whom it persisted. Chloride was not analysed separately, so the hyperchloraemic contribution attributable to resuscitation fluid could not be quantified [16]. Serum albumin was not incorporated, which may have obscured occult unmeasured anions in hypoalbuminaemic patients [17]. Finally, the sample of 100 from a single centre limits precision and external applicability.

In this prospective cohort of 100 critically ill patients with septic shock, all four admission arterial blood gas acid–base parameters differed significantly between survivors and non-survivors. Overall in-hospital mortality was 53%, rising to 89.6% among patients presenting with a blood pH below 7.20 and to 85.7% among those with a base deficit exceeding 11 mEq/L. Non-survivors demonstrated a mean pH of 7.13 against 7.31 in survivors, a mean bicarbonate of 12.9 against 19.3 mEq/L, a mean base excess of −14.2 against −4.0 mEq/L, and profound compensatory hypocapnia with a mean pCO₂ of 18.8 against 29.8 mmHg. Base excess showed the widest separation between outcome groups and, because it isolates the metabolic component of the disturbance from respiratory compensation, offers the most interpretable single index of severity. Arterial blood gas analysis is inexpensive, rapidly available and already routine in the intensive care unit; the present data support its systematic use for early risk stratification in septic shock, so that patients presenting with severe metabolic acidosis are recognised immediately as a group requiring the most aggressive and closely monitored resuscitation. Serial rather than single measurement should be preferred wherever feasible.

1. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):801-10.
2. Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of sepsis in the United States from 1979 through 2000. N Engl J Med. 2003;348(16):1546-54.
3. Vincent JL, Rello J, Marshall J, Silva E, Anzueto A, Martin CD, et al. International study of the prevalence and outcomes of infection in intensive care units. JAMA. 2009;302(21):2323-9.
4. Kraut JA, Madias NE. Metabolic acidosis: pathophysiology, diagnosis and management. Nat Rev Nephrol. 2010;6(5):274-85.
5. De Backer D. Lactic acidosis. Minerva Anestesiol. 2003;69(4):281-4.
6. Kellum JA, Bellomo R, Kramer DJ, Pinsky MR. Etiology of metabolic acidosis during saline resuscitation in endotoxemia. Shock. 1998;9(5):364-8.
7. Gunnerson KJ. Clinical review: the meaning of acid-base abnormalities in the intensive care unit. Crit Care. 2005;9(5):508-16.
8. Smith I, Kumar P, Molloy S, Rhodes A, Newman PJ, Grounds RM, et al. Base excess and lactate as prognostic indicators for patients admitted to intensive care. Intensive Care Med. 2001;27(1):74-83.
9. Kaplan LJ, Kellum JA. Initial pH, base deficit, lactate, anion gap, strong ion difference, and strong ion gap predict outcome from major vascular injury. Crit Care Med. 2004;32(5):1120-4.
10. Dunham CM, Siegel JH, Weireter L, Fabian M, Goodarzi S, Guadalupi P, et al. Oxygen debt and metabolic acidemia as quantitative predictors of mortality and the severity of the ischemic insult in hemorrhagic shock. Crit Care Med. 1991;19(2):231-43.
11. Jung B, Rimmele T, Le Goff C, Chanques G, Corne P, Jonquet O, et al. Severe metabolic or mixed acidemia on intensive care unit admission: incidence, prognosis and administration of buffer therapy. A prospective, multiple-center study. Crit Care. 2011;15(5):R238.
12. Davis JW, Parks SN, Kaups KL, Gladen HE, O'Donnell-Nicol S. Admission base deficit predicts transfusion requirements and risk of complications. J Trauma. 1996;41(5):769-74.
13. Kiran HS, Anil GD, Sudharshana Murthy KA, Basavana Gowdappa H. Severe metabolic acidosis in critically ill patients and its impact on the outcome: a prospective observational study. Int J Sci Study. 2015;3(6):39-44.
14. Noritomi DT, Soriano FG, Kellum JA, Cappi SB, Biselli PJ, Libório AB, et al. Metabolic acidosis in patients with severe sepsis and septic shock: a longitudinal quantitative study. Crit Care Med. 2009;37(10):2733-9.
15. Mecher C, Rackow EC, Astiz ME, Weil MH. Unaccounted for anion in metabolic acidosis during severe sepsis in humans. Crit Care Med. 1991;19(5):705-11.
16. Scheingraber S, Rehm M, Sehmisch C, Finsterer U. Rapid saline infusion produces hyperchloremic acidosis in patients undergoing gynecologic surgery. Anesthesiology. 1999;90(5):1265-70.
17. Figge J, Jabor A, Kazda A, Fencl V. Anion gap and hypoalbuminemia. Crit Care Med. 1998;26(11):1807-10.
18. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M, Ferrer R, et al. Surviving Sepsis Campaign: international guidelines for management of sepsis and septic shock: 2016. Crit Care Med. 2017;45(3):486-552.
19. Levy MM, Evans LE, Rhodes A. The Surviving Sepsis Campaign Bundle: 2018 update. Crit Care Med. 2018;46(6):997-1000.
20. Cusack RJ, Rhodes A, Lochhead P, Jordan B, Perry S, Ball JA, et al. The strong ion gap does not have prognostic value in critically ill patients in a mixed medical/surgical adult ICU. Intensive Care Med. 2002;28(7):864-9.