Acid-base disorders in sick goats and their association with mortality: A simplified strong ion difference approach.

OBJECTIVES: To investigate the acid-base status of sick goats using the simplified strong ion difference (sSID) approach, to establish the quantitative contribution of sSID variables to changes in blood pH and HCO3 - and to determine whether clinical, acid-base, and biochemical variables on admission are associated with the mortality of sick goats. ANIMALS: One hundred forty-three sick goats.
METHODS: Retrospective study. Calculated sSID variables included SID using 6 electrolytes unmeasured strong ions (USI) and the total nonvolatile buffer ion concentration in plasma (Atot ). The relationship between measured blood pH and HCO3 - , and the sSID variables was examined using forward stepwise linear regression. Cox proportional hazard models were constructed to assess associations between potential predictor variables and mortality of goats during hospitalization.
RESULTS: Hypocapnia, hypokalemia, hyperchloremia, hyperlactatemia, and hyperproteinemia were common abnormalities identified in sick goats. Respiratory alkalosis, strong ion acidosis, and Atot acidosis were acid-base disorders frequently encountered in sick goats. In sick goats, the sSID variables explained 97% and 100% of the changes in blood pH and HCO3 - , respectively. The results indicated that changes in the respiratory rate (<16 respirations per minute), USI, and pH at admission were associated with increased hazard of hospital mortality in sick goats. CONCLUSIONS AND CLINICAL IMPORTANCE: The sSID approach is a useful methodology to quantify acid-base disorders in goats and to determine the mechanisms of their development. Clinicians should consider calculation of USI in sick goats as part of the battery of information required to establish prognosis.

total net negative charge of plasma proteins

total plasma concentration of nonvolatile weak acids

ß‐hydroxybutyrate

bicarbonate

Henderson‐Hasselbalch approach

L‐lactate

arterial partial carbon dioxide pressure

venous partial carbon dioxide pressure

propylene glycol

strong ion difference

strong ion difference measured using 6 electrolytes

measured strong ion difference

simplified strong ion difference approach

total plasma protein

unmeasured strong ions

INTRODUCTION

Sick goats often are presented to tertiary hospitals with critical conditions such as acute diarrhea, pneumonia, gastrointestinal parasitism, pregnancy toxemia, mastitis, and urinary diseases such as urolithiasis. Regardless of presentation, these clinical abnormalities frequently result in marked acid‐base, fluid, electrolyte, and plasma protein disorders. The mechanisms contributing to acid‐base disorders in sick goats have not been extensively investigated. The few published studies focused on analysis of acid‐base disorders in goats with pregnancy toxemia , and urolithiasis using the traditional Henderson‐Hasselbach (H‐H) approach. In recent decades, different studies in sick horses, , , , and cattle , , have demonstrated the quantitative contribution of strong ions, plasma proteins, and PCO2 to the changes in plasma H+ activity and HCO3 − concentration by using the simplified strong ion difference (sSID) approach. , , The sSID proposes that 3 independent variables modify the plasma H+ activity. These 3 variables are the arterial partial carbon dioxide pressure (PaCO2), the strong ion difference (SID), and the total weak acid concentration (Atot −). The PaCO2 is regulated by the lungs and influences plasma H+ activity in a linear fashion. , Strong ion difference accounts for the contribution of the strong cations (Na+, K+, Ca2+, Mg2+) and strong anions (Cl−, L‐lactate− [L‐lac−]), D‐lactate−, uremic acids, ketoacids, and unmeasured strong ions (USI) to the changes in plasma pH and HCO3 − concentration. , The Atot − includes important nonvolatile weak buffers (mostly proteins and phosphate) that exert an independent effect on plasma H+ activity. , , The sSID approach has been used to determine the acid‐base balance of healthy goats. However, no published studies have evaluated acid‐base disorders in sick goats using the sSID.

Increased concentrations of USI other than L‐lac− have been reported in critically ill human , and bovine patients. , , The source of these anions is unclear, but USI have been detected in human patients with sepsis, , decreased tissue perfusion, and kidney and liver injury. , , Increased concentrations of USI also were reported in calves with diarrhea, and the concentrations correlated with plasma concentrations of lipopolysaccharides. Higher concentrations of USI also were associated with increased risk of mortality in human , and bovine neonates. Currently, no information is available regarding the predictive value for mortality of admission concentrations of USI in sick goats.

Our objectives were to investigate the acid‐base status of sick goats using the sSID approach, to establish the quantitative contribution of the measured strong ion difference (SIDm), PCO2, USI, and total plasma proteins (TPP) to changes of blood pH and HCO3 − concentration. Our study also aimed to investigate whether clinical, acid‐base, and biochemical variables, especially USI, on admission are associated with the mortality of sick goats. We hypothesized that sSID variables explain the majority of the changes in blood pH and HCO3 − concentration of sick goats, and that admission concentrations of USI and L‐lac− are associated with mortality of sick goats.

MATERIALS AND METHODS

Study population and data collection

Medical records of all goats admitted at the Large Animal Hospital of the University of Florida, College of Veterinary Medicine, between January 2017 and September 2019 were reviewed. Goats were included in the study if they were >3 months of age, and if they had blood gas and TPP concentrations determined on admission before any treatment was instituted.

The following data were extracted from each record: age, breed, sex, heart rate (HR, beats per minute, bpm), respiratory rate (RR, respirations per minute, rpm), rectal temperature (T, °F), packed cell volume (PCV, %), TPP (g/dL), blood pH, venous partial pressure of carbon dioxide (PvCO2, mm Hg), bicarbonate concentration (HCO3 −, mmol/L), electrolyte concentrations (Na+, K+, Mg2+, Ca2+, Cl−, L‐lac−, all in mmol/L), anion gap (AG, mEq/L), blood glucose concentration (mg/dL), blood ß‐hydroxybutyrate (BHB) concentration, and blood creatinine concentration (mg/dL). All blood gas analyses were corrected for the patient's rectal temperature using a standard method employed by the blood gas analyzer.

Sample collection and measurement techniques

Venous blood samples were collected in plastic blood collection tubes containing sodium heparin and obtained from the jugular vein at the time of admission. Venous blood gas analysis (VBGA), electrolyte, glucose, and creatinine concentrations were determined using a blood gas analyzer (Nova Biomedical Stat Profile, Novabio, Waltham, Massachusetts). The Na+, K+, Ca2+, Mg2+, and Cl− concentrations were measured using ion‐selective electrode technology based on direct potentiometry. Blood pH and PvCO2 were measured using ion‐selective electrodes based on direct and indirect potentiometry, respectively, both based on the Nernst equation. The H‐H equation was used to calculated blood HCO3 − concentration (mmol/L): [HCO3 −] = S × PCO2 × 10(pH – pK1) with experimentally determined values for S (0.0307 [mmol/L]/mm Hg) and pK1 (6.095 at [NaCl] = 0.16 mmol/L) in plasma. The glucose, creatinine and L‐lac− concentrations were determined using an enzymatic‐amperometric technique. The PCV was measured using a microhematocrit tube and TPP concentration was determined using refractometry (Cole‐Parmer, Vernon Hills, Illinois). The BHB concentration was also determined using an enzymatic‐amperometric technique (Abbott Diabetes Care, Alameda, California).

Calculations

The sSID variables were calculated using the formulas previously described for cattle as follows: strong ion difference (SIDm, mEq/L) was calculated using the concentrations of 6 strong ions as: SID6 = (Na+ + K+ + Ca2+ + Mg2+) – (Cl− + L‐lac−). The Atot (mmol/L) was calculated as: Atot = [3.43 × (TPP)], where TPP is in grams per deciliter. The USI (mEql/L) concentration was calculated as: USI = SID6 – HCO3 − – ([Atot −]/(1 + 10(pKa‐pH))), where pKa (7.08) is the effective dissociation constant of bovine plasma weak acids. Species‐specific values for Ka and Atot have not been experimentally determined for goats, therefore values experimentally determined for calves were used.

Acid‐base disorders definition

Acid‐base disorders were diagnosed when the H‐H and sSID variables were outside of the reference ranges (Table 1). , ,

TABLE 1

Definition of acid‐base disorders using the Henderson‐Hasselbalch (H‐H) and the simplified strong ion difference (sSID) approach

Type of disorder	Approach	Variable	Acidosis	Alkalosis
Respiratory	H‐H and sSID	PvCO2 [34‐48 mm Hg]	↑ PvCO2	↓ PvCO2
Metabolic	H‐H	HCO3 − [20‐30 mmol/L]	↓ HCO3 −	↑ HCO3 −
		AG <24 mEq/L	↑ AG	N/A
	sSID	SID6 [38‐46 mEq/L]	↓ SID6	↑ SID6
		Atot [19‐25 mmol/L]	↑ Atot	↓ Atot
		USI [−3 to 3 mEq/L]	↑USI	N/A

Note: Modified from References 10, 66.

Abbreviations: AG, anion gap; Atot total plasma concentration of nonvolatile weak acids; HCO3 −, bicarbonate; PvCO2, partial venous carbon dioxide pressure; SID, strong ion difference measured; USI unmeasured strong ions.

Statistical analysis

Descriptive analyses were performed on all variables. A Kolmogorov‐Smirnov test was used to assess normality of the data. Depending on the distribution, mean and standard deviation or median and range were calculated for each of the variables. Differences among goats with different clinical diagnoses were evaluated using the Steel‐Dwass test for nonparametric multiple comparisons.

The relationship between measured blood pH and HCO3 −, and the sSID variables (SID6, USI, PvCO2, and Atot) was examined using a forward stepwise linear regression. , , The quantitative contribution of each variable was evaluated by order of entry into the model and change of the model R 2 value. , , The partial R or the potential contribution of the added variable for explaining the variation within the model was assessed by subtracting the R of the previous model from R of the current model. , , Residuals plots were examined to investigate the assumptions of the forward stepwise regression procedure. The regression results were confirmed using backward elimination. Evaluation of multicollinearity was carried out using the variance inflation factor of all significant predictors.

To determine which clinicopathological variables were associated with the outcome of sick goats (discharged or euthanized or died), Cox proportional hazard models were constructed to assess the potential predictor variables and the outcome of interest. Days from admission to the hospital until discharged or died or euthanized was used as the time variable. Variables included into the analysis were sex, HR, RR, rectal T (T), creatinine, glucose, PCV, TPP, Na+, K+, Cl−, Ca2+, Mg2+, L‐lac−, pH, PvCO2, HCO3 −, AG, SID6, and USI. The sex of the goats was categorized as male, female, or wether. Martingale residuals were plotted against the continuous variables to assess the linearity of each continuous predictor variable offered to the Cox proportional hazard models. If the variable failed to meet the linearity assumption, it was categorized as follows based on biologically relevant ranges , : HR was categorized as <70, 70 to 80, and >80 bpm; RR as <16, 16 to 30, and >30 rpm; T as <101.3°F, 101.3°F to 103.5°F, and >103.5°F; blood creatinine concentration as ≤0.8 and >0.8 mg/dL; blood HCO3 − concentration as <20, 20 to 30, and >30 mmol/L; AG as ≤24 or >24 mmol/L; K+ as <4.2, 4.2 to 6, and >6 mmol/L; and Cl− as <98 mmol/L, 98 to 110, and >110 mmol/L. Collinearity among variables was tested using Pearson correlation coefficients and if r > 0.6, only 1 variable was retained in the model. Variables with moderate statistical associations (P < .2) with the outcome of interest in univariable models subsequently were included in the multivariable analyses. A manual backward stepwise multivariable regression analysis was used for the different models. Variables were retained in the multivariable model if P < .05 or if the effect of removing the variable resulted in a change of at least 20% in the coefficient of a significant variable, indicating potential confounding. Two‐way interactions were evaluated between variables based on evidence from the literature and retained in the model if significant. Model fit was assessed using the test of proportional hazards for the Cox proportional hazards models. Harrell's C concordance statistic for each survival model was calculated to evaluate overall predictive ability. Harrell's C concordance statistic is calculated by taking all possible pairs of subjects consisting of 1 subject that experienced the event of interest and 1 subject that did not experience the event of interest. The statistic is the proportion of such pairs in which the subject that experienced the event had a higher predicted probability of experiencing the event than the subject that did not experience the event. A value of C = 0.5 corresponds to non‐informative prediction, whereas, C = 1.0 corresponds to perfect prediction. P‐values <.05 were considered significant.

For continuous variables that were significant in the final survival models, a cutoff was determined by receiver operator characteristics curve (ROC) analysis. The optimal cutoff point was based on the Youden's index and the corresponding sensitivity, specificity, and area under the curve were calculated for this cutoff point. Statistical analyses were performed using Minitab 19 (Minitab, LLC, State College, Pennsylvania) and STATA 14 (StataCorp LLC, College Station, Texas) statistical software.

RESULTS

Study population

One‐hundred forty‐three goats met the inclusion criteria. Seventy‐five (53%) goats were presented in 2017, 45 (31%) in 2018, and 23 (16%) in 2019. The age was recorded for 140 goats; median age on admission was 2 years (range, 3 months to 14 years). Of the 143 goats, 101 (71%) were females, 18 (12%) were males, and 24 (17%) were wethers. Breed distribution was similar to that of the general hospital population, with Nigerian Dwarf (n = 36, 25%) being the most commonly represented. Remaining breeds included La Mancha (n = 24, 17%), Nubian (n = 20, 14%), Oberhasli (n = 17, 12%), Boer (n = 8, 6%), Pygmy (n =7, 5%), Toggenburg (n = 5, 3%), Alpine (n = 2, 1%), and other breeds (n = 24, 17%). The values (mean, SD, median and range) for HR, RR, T, VBGA, electrolyte concentrations, TPP, and the H‐H and sSID variables for all 143 goats are presented in Supplementary Table 1.

Electrolyte, biochemical and acid‐base abnormalities in sick goats

Nineteen (13%) goats were diagnosed with acute diarrhea, 19 (13%) with urolithiasis, 16 (11%) were presented for dystocia, 14 (10%) with haemonchosis, 13 (9%) with pregnancy toxemia, 10 (7%) with pneumonia, and 7 (5%) with clinical mastitis. Forty‐five (32%) goats were diagnosed with other diseases including azalea toxicity (n = 4, 3%), hypocalcemia (n = 3, 2%), simple indigestion (n = 3, 2%), vagal indigestion (n = 3, 2%), neoplasia (n = 3, 2%), ruminal foreign body (n = 2, 1%), and peritonitis (n = 2, 1%).

The results for blood gas analysis, electrolyte concentrations, TPP, H‐H, and sSID variables for 98 goats with selected diseases are presented in Table 2. Goats suffering from haemonchosis had significantly lower PCV (median, 8%; range, 7%‐18%) and TPP (median, 4.4 g/dL; range, 3.5‐7.0 g/dL) than did goats from other groups (P < .05). Blood glucose concentration was higher (median, 193 mg/dL; range, 90‐325 mg/dL) in goats with diarrhea than in other groups (P < .05), except the urolithiasis group. Goats with urolithiasis had blood creatinine concentration (median, 3.4 mg/dL; range, 0.6‐12.5 mg/dL) higher than did the other groups (P < .05), except the clinical mastitis group, whereas goats with dystocia had higher concentration of Cl− (median, 115 mmol/L; range, 108‐119 mmol/L) than did the other groups (P < .05). The Atot was signifficantly lower in goats with haemonchosis (median, 15 mmol/L; range, 12‐23 mmol/L) than other groups (P < .05, for all comparisons). Goats with pregnancy toxemia had a higher concentration of USI (median, 3.8 mEq/L; range, −2.4 to 14 mEq/L) than did goats with acute diarrhea, urolithiasis, haemonchosis, and clinical mastitis (P < .05). Concentrations of BHB (mmol/L) were only measured in goats with pregnancy toxemia with a median of 3 mmol/L (range, 1.2‐8 mmol/L; Table 2).

TABLE 2

Admission results of blood gas, electrolytes, acid‐base and selected laboratory variables of 98 sick goats with selected diseases

Variable	Acute diarrhea	Pneumonia	Urolithiasis	Haemonchus infection	Pregnancy toxemia	Clinical mastitis	Dystocia
Cases	19	10	19	14	13	7	16
PCV (%)	37be 29 to 55	30bd 29 to 60	29cd 19 to 47	8a 7 to 18	25cd 12 to 42	32bcd 25 to 39	30cd 23 to 52
TPP (g/dL)	7.1b 6.4 to 8.2	6.5bc 4.8 to 9.4	6.3c 4 to 7.3	4.4a 3.5 to 7	6.6bc 5.5 to 7.5	6.5bc 6.2 to 8	6.5bc 5 to 8
Glucose (mg/dL)	193a 90 to 325	113bc 34 to 284	131ac 67 to 380	105bc 23 to 190	95bc 51 to 160	62bc 37 to 99	119bc 67 to 284
Creatinine (mg/dL)	1.1bc 0.4 to 3	0.9bc 0.4 to 1.7	3.4a 0.6 to 12	0.75bc 0.5 to 1.2	1bc 0.7 to 2	1ac 0.6 to 3	1bc 0.6 to 2.4
Na+ (mmol/L)	141a 134 to 151	144a 137 to 146	143ab 131 to 152	138a 135 to 148	143ab 136 to 150	141ab 135 to 147	145b 141 to 152
K+ (mmol/L)	3.6b 2.4 to 4.6	4ab 2 to 5.2	4.6a 2.2 to 10	3.5b 2 to 4.6	3.8b 2.4 to 4.5	4ab 2.8 to 4.6	4.1ab 2.8 to 4.6
Cl− (mmol/L)	108ad 96 to 112	113abc 87 to 119	109ab 100 to 119	111bc 94 to 118	111bc 108 to 118	112bc 108 115	115c 108 119
Ca2+ (mmol/L)	1.1a 0.9 to 1.2	1.1a 0.8 to 1.4	1.2a 0.8 to 1.6	1.1a 1 to 1.2	1.1a 11 to 1.3	1.1a 1 to 1.2	1.1a 1 to 1.2
Mg2+ (mmol/L)	0.5a 0.4 to 1	0.5a 0.3 to 0.7	0.6a 0.4 to 1.3	0.5a 0.4 to 0.7	0.5a 0.4 to 0.6	0.5a 0.4 to 0.6	0.5a 0.4 to 0.6
L‐lac− (mmol/L)	3.8a 0.8 to 7.2	2.1bc 0.4 to 7.2	1.5bc 0.5 to 8.6	2.8ac 1 to 10	1.4ac 0.3 to 7	2.1ac 0.3 to 5.6	2.6ac 0.4 to 7.2
pH	7.44ab 7.30 to 7.50	7.46ab 7.38 to 7.52	7.47b 7.36 to 7.55	7.46b 7.37 to 7.53	7.37a 7.18 to 7.52	7.46ab 7.38 to 7.48	7.468ab 7.39 to 7.51
PvCO2 (mm Hg)	31a 22 to 42	26ab 21 to 46	29ab 20 to 60	26ab 17 to 43	24b 18 to 32	25ab 16 to 30	25b 21 to 31
HCO3 − (mmol/L)	20bc 14 to 30	19bd 14 to 34	21bc 15 to 33	19bd 11 to 33	16ad 7.7 to 21	18bd 11 to 22	19bd 14 to 26
AG (mEq/L)	16cd 13 to 22	16abc 5.3 to 25	15b 8 to 19	12c 7 to 17	19a 11 to 27	15abc 12 to 19	17abd 13 to 19
SID6 (mEq/L)	35ab 30 to 44	35ab 29 to 41	38a 26 to 49	29b 19 to 46	33ab 28 to 40	30ab 23 to 37	30ab 23 to 21
Atot (mmol/L)	24b 22 to 28	22bc 16 to 32	21c 14 to 25	15a 12 to 23	22bc 29 to 26	22bc 21 to 27	22bc 17 to 28
USI (mEq/L)	−2.9b −5 to 2.1	−0.8ab −4.6 to 6.6	−1.4b −3.5 to 3.3	−1.9b −3.8 to 1.6	3.8a −2.8 to 14	−3.5b −5.1 to 1.6	−0.8ab −3.3 to 6.6

Notes: Data presented as median and lower and higher range. Different letters within a row indicated a statistically significant difference (P < .05). P‐values were obtained using the Steel‐Dwass test for nonparametric multiple comparisons.

Abbreviations: AG, anion gap; Atot, total plasma concentration of nonvolatile weak acids; HCO3 −, bicarbonate; L‐lac−, L‐lactate; PCV, packet cell volume; PvCO2, venous partial carbon dioxide pressure; SID, strong ion difference; TPP, total plasma proteins; USI, unmeasured strong ions.

The electrolyte, TPP, and acid‐base disorders detected in 143 sick goats are presented in Table 3. Overall, the most common electrolyte abnormalities in sick goats were hyper‐L‐lactatemia (n = 96, 67%), hypokalemia (n = 89, 62%), and hyperchloremia (n = 76, 53%). Hyponatremia was present in 6 animals (4%) and hypernatremia was not detected in any of the goats. Hypoproteinemia frequently was diagnosed in goats with haemonchosis (n = 11, 78%).

TABLE 3

Acid‐base, electrolyte, and total plasma protein disorders detected on admission in 143 sick goats with selected diseases

	Acute diarrhea	Pneumo	Urolith	Haemon	Preg. Tox.	Mastitis	Dystocia	Others
Cases (n)	19	10	19	14	13	7	16	45
Na+
Hyponatremia	0 (0%)	1 (10%)	2 (11%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	3 (7%)
Cl−
Hypochloremia	2 (11%)	2 (20%)	0 (0%)	1 (7%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Hyperchloremia	2 (11%)	5 (50%)	8 (42%)	8 (57%)	7 (54%)	6 (86%)	15 (94%)	25 (56%)
K+
Hypokalemia	15 (79%)	7 (70%)	4 (21%)	11 (79%)	10 (77%)	4 (57%)	9 (56%)	29 (64%)
Hyperkalemia	0 (0%)	0 (0%)	3 (16%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
L‐Lactate−
Hyperlactatemia	18 (95%)	4 (40%)	10 (53%)	13 (93%)	7 (54%)	5 (71%)	12 (75%)	27 (60%)
TPP
Hypoprotein	0 (0%)	2 (20%)	6 (32%)	11 (79%)	5 (38%)	0 (0%)	6 (38%)	13 (29%)
Hyperprotein	1 (5%)	2 (20%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	1 (6%)	4 (9%)
pH
Acidemia	0 (0%)	0 (0%)	0 (0%)	0 (0%)	2 (15%)	0 (0%)	0 (0%)	1 (2%)
Alkalemia	2 (11%)	2 (20%)	7 (37%)	2 (14%)	1 (8%)	0 (0%)	4 (25%)	10 (22%)
PvCO2
Resp. acidosis	0 (0%)	0 (%)	1 (5%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Resp. alkalosis	16 (84%)	8 (80%)	14 (74%)	12 (86%)	13 (100%)	7 (100%)	16 (100%)	39 (87%)
HCO3 −
Met. acidosis	5 (26%)	6 (60%)	4 (21%)	8 (57%)	11 (85%)	6 (86%)	10 (63%)	18 (40%)
Met. alkalosis	0 (0%)	1 (10%)	3 (16%)	1 (7%)	0 (0%)	0 (0%)	0 (0%)	1 (2%)
AG
Acidosis	0 (0%)	1 (10%)	0 (0%)	0 (0%)	2 (15%)	0 (0%)	0 (0%)	2 (4%)
SID6
Acidosis	15 (79%)	9 (90%)	11 (58%)	13 (93%)	12 (92%)	7 (100%)	14 (88%)	36 (80%)
Alkalosis	0 (0%)	0 (0%)	0 (0%)	1 (7%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Atot
Acidosis	7 (37%)	4 (40%)	1 (5%)	0 (0%)	3 (23%)	3 (43%)	2 (13%)	12 (27%)
Alkalosis	0 (0%)	2 (20%)	2 (10%)	11 (71%)	2 (15%)	0 (0%)	4 (26%)	7 (15%)
USI
Acidosis	0 (0%)	0 (0%)	1 (5%)	0 (%)	8 (62%)	0 (0%)	1 (6%)	4 (9%)

Abbreviations: AG, anion gap; Atot, total plasma concentration of nonvolatile weak acids; Haemon, haemonchus infection; HCO3 −, bicarbonate; Hyperlactate, hyperlactatemia; Hyperprotein, hyperproteinemia; Hypoprotein, hyporproteinemia; Pneumo, pneumonia; PvCO2, venous partial carbon dioxide pressure; SID, strong ion difference; TPP, total plasma proteins; USI, unmeasured strong ions.

Based on the H‐H approach, the most common acid‐base disorders in sick goats were respiratory alkalosis (n = 125, 87%) and metabolic acidosis (n = 68, 48%). Anion gap acidosis was detected in only 2 (1%) goats. When using the sSID approach, the most common acid‐base disorders were respiratory alkalosis (n = 125, 87%) and SID6 acidosis (n = 117, 82%). A USI acidosis was diagnosed in 14 (10%) goats, whereas Atot alkalosis was present in 17 (12%), and Atot acidosis in 84 (59%) goats.

The most common sSID acid‐base disorders in goats with acute diarrhea, pneumonia, urolithiasis, mastitis, and dystocia were respiratory alkalosis and SID6 acidosis. In goats with pregnancy toxemia, respiratory alkalosis, SID6 acidosis and USI acidosis frequently were detected (n = 8, 62%). Hypoproteinemic (Atot) alkalosis and hyperproteinemic (Atot) acidosis were commonly detected in goats with haemonchosis (n = 11, 72%) and acute diarrhea (n = 7, 37%), respectively.

variables and blood pH and HCO

The results of forward stepwise regression using venous blood pH and calculated HCO3 − concentrations as dependent variables showed that, in sick goats, PvCO2 was the most important variable associated with variations in the measured blood pH and calculated HCO3 −. The SID6, USI, and Atot also were associated with changes in the measured blood pH and HCO3 − (Table 4).

TABLE 4

Results of the stepwise linear regression of measured blood pH and calculated HCO3 − as dependent variables versus jugular venous values of the simplified strong ion difference variables of 143 sick goats

	Order of entry	Variable	Partial R 2	Model R 2	VIF
pH	1	SID6	9.5	9.5	2.46
	2	USI	13.5	23	1.60
	3	Atot	24	47	1.57
	4	PvCO2	50.3	97.3	1.94
HCO3 −	1	PvCO2	59.8	59.8	2.6
	2	SID6	12.1	71.9	1.59
	3	Atot	8.4	80.3	1.55
	4	USI	19.6	99.9	2.09

Note: Modified from Reference 11.

Abbreviations: Atot total plasma concentration of nonvolatile weak acids; PvCO2 partial venous carbon dioxide pressure; SID6 Measured strong ion difference measured using 6 electrolytes; USI, unmeasured strong ions.

Factors associated with mortality in hospitalized sick goats

One‐hundred sixteen (81%) goats survived to hospital discharge whereas 27 (19%) did not. Six (22%) of the 27 nonsurviving goats died during hospitalization and 21 (80%) were euthanized. Eight goats were euthanized (n = 7) or died (n = 1) on day 1 of hospitalization. The goat that died on day 1 of hospitalization was being treated for haemonchosis by blood transfusion. Diagnoses of euthanized goats on day 1 of hospitalization included end stage liver disease (n = 2), severe trauma from dog attack (n = 1), urolithiasis (n = 1), diarrhea (n = 1), and chronic bronchopneumonia (n = 1). Diagnoses of the goats that died during hospitalization (n = 6) included haemonchosis (n = 2), peritonitis (n = 1), meningitis (n = 1), pregnancy toxemia (n = 1), and bronchopneumonia (n = 1). Diagnoses of the goats that were euthanized during hospitalization (n = 20) included neoplasia (n = 3), end stage liver disease (n = 2), emaciation (n = 2) pregnancy toxemia (n = 2), urolithiasis (n = 2), diarrhea (n = 1), nephrotic syndrome (n = 1), severe trauma from dog attack (n = 1), caprine arthritis‐encephalitis (n = 1), vagal indigestion (n = 1), toxic mastitis (n = 1), chronic pneumonia (n = 1), listeriosis (n = 1), and dystocia (n = 1). The time of hospitalization (median and range) for surviving goats was 3 days (0‐20 days) and for nonsurviving goats was 2 days (0‐9 days; P = .17).

Admission values (median and range) of blood gas, electrolytes, acid‐base and selected laboratory variables of surviving and nonsurviving sick goats are presented in Supplementary Table 2. Variables associated with mortality in the univariable analysis were HR <80 bpm, RR <16 rpm, HCO3 − <20 mmol/L, glucose, blood pH, SID6, and USI. Blood pH and USI were highly correlated (Supplementary Table 3) and therefore 2 different multivariable Cox proportional hazard models were built using each of those variables (Table 5). The first multivariable Cox proportional model assessing the association between selected variables and survival included RR <16 rpm and USI (Table 5). The model correctly ordered survival times for pairs of goats 68% of the time (Harrell's C concordance statistic, 0.68). The optimal cut point for USI was determined to be −0.49 mmol/L based on Youden's index. At this cut point, the sensitivity, specificity and area under the ROC curve for predicting nonsurvival was 52%, 69%, and 0.60, respectively. The second multivariable Cox proportional model using blood pH for predicting hospital mortality included the variables RR <16 rpm and blood pH. The SID6 was retained in the latter model because it changed the coefficient of pH by >20% indicating a potential confounding effect (Table 5). The Harrell's C concordance statistic was 0.66, indicating the model could correctly order survival times for pairs of goats 66% of the time based on pH, RR, and SID6. The optimal cut point for pH was determined to be 7.45 mmol/L. At this cut point, the sensitivity, specificity, and area under the ROC curve for predicting nonsurvival was 56%, 53%, and 0.55, respectively.

TABLE 5

Results of the multivariable Cox proportional hazard models assessing the association between selected admission clinical and laboratory variables and outcome of 143 sick goats

Variable		HzR	SE	95% CI	P value
Model 1
USI (mEq/L)	Every 1 mEq/L USI unit increase	1.13	0.05	1.04‐1.22	.004
RR (rpm)	Normal (16‐30)	Referent
	>30	1.14	0.55	0.45‐2.93	.78
	<16	9.62	8.33	1.76‐52	.009
Model 2
RR (rpm)	Normal (16‐30)	Referent
	>30	1.09	0.52	0.43‐2.8	.86
	<16	7.50	6.47	1.38‐41	.02
Blood pH	Every 0.01 pH unit increase	0.93	0.03	0.89‐0.98	.01
SID6 (mEq/L)	Every 1 mEq/L SID6 unit increase	1.07	0.04	0.99‐1.16	.08

Abbreviations: CI, confidence interval; HzR, hazard ratio; RR, respiratory rate; SID, strong ion difference.

DISCUSSION

We found that hypocapnia, hypokalemia, hyperchloremia, hyperlactatemia, and hyperproteinemia were common abnormalities identified in sick goats, and that respiratory alkalosis, SID6 acidosis, and Atot acidosis were acid‐base disorders frequently encountered in hospitalized goats. The results also indicated that changes in RR (<16 rpm), USI, and blood pH on admission were associated with hospital mortality in sick goats. According to our results, sSID variables accounted for most of the changes in blood pH and HCO3 −. The sSID variables explained 97% and 100% of the changes in blood pH and HCO3 −, respectively. Similar results were reported in sick calves , and foals. In calves, with or without diarrhea, sSID contributed to 85% to 93% of the change in blood pH and 98% of the change in HCO3 −, whereas in foals, sSID variables explained 83% to 94% of the changes in blood pH.

Our findings are in agreement with previous research in cattle and horses showing that the sSID approach provides an accurate evaluation of acid‐base homeostasis because it was able to detect complex mixed alteration in SID6, PvCO2, USI, and Atot. The H‐H approach states that primary respiratory and metabolic disturbances compensate for each other. For example, a respiratory disorder will be compensated by renal excretion or reabsorption of HCO3, whereas a metabolic disorder will be rapidly compensated by alteration in pCO2 by the respiratory system. Describing primary and compensatory responses to acid‐base disturbances using HCO3 − is qualitative and offers little information of the physiological status of the patient. The sSID approach determines quantitatively how a primary acid‐base disturbance in PCO2, Atot or SID6 produces an effect on the other 2 variables to initiate a compensatory response. The sSID approach states that SIDm (SID6 in our study) has a normal value of approximately 40 mmol/L in a single compartment system where Atot and PCO2 remain constant at normal values. , , However, a codependency between SIDm, PCO2, and Atot exists when the system is defined as whole body acid‐base balance. Thus, the respiratory alkalosis (hypocapnia) observed in the majority of the goats in our study could be a primary physiological (eg, pregnancy, pain, and stress) or a pathological (eg, pneumonia, sepsis, and anemia) alteration, resulting in increased alveolar ventilation. Hypocapnia as a primary alteration leads to changes in all body H+ activity including that in plasma, intracellular and interstitial fluids. An acute, but especially chronic, change in PCO2 stimulates alterations in SIDm that are adjusted mostly by alteration in the Na+ and Cl− ratio by kidney regulation of Cl−. , , , , It is likely that during respiratory alkalosis in sick goats, the renal excretion of Cl− decreases, resulting in hyperchloremia, and subsequently lowering SIDm. Therefore, in cases of primary respiratory alkalosis, SID6 acidosis should not be interpret as a mixed acid‐base disorder, but rather as a new “normal” SID6 value.

In a subset of goats enrolled in our study, respiratory alkalosis could be a compensatory response to the changes in blood pH because of SID6 acidosis, USI acidosis, or Atot acidosis. Changes in SID6 could result from a high [Cl−] relative to [Na+] (eg, acute diarrhea, mastitis), increased [L‐lac−] (eg, haemonchosis, mastitis, and dystocia) or increased USI (eg, BHB in pregnancy toxemia). The effect of high [Cl−] relative to [Na+] on blood pH and HCO3 − has been demonstrated in children, calves, , , horses with diarrhea, critically ill calves and foals with sepsis. In fact, in diarrheic horses, fecal fluid [Cl−] losses were lower than those of [Na+] and urinary excretion of [Cl−] was decreased. Plasma [Cl−] then increases, lowering SIDm, and causing SID acidosis. The reasons for the hyperchloremia in goats with mastitis and pneumonia are not clear, but those goats usually presented to our institution with signs of sepsis. Similarly, in humans with severe sepsis and septic shock, hyperchloremic metabolic acidosis is diagnosed frequently.

Hyperlactatemia was diagnosed commonly in sick goats, especially diarrheic goats and those with haemonchosis. Several studies in humans, , cattle, , , , horses, , and dogs have reported hyperlactatemia to be present in the majority of critically ill patients. L‐lactate is a strong anion and its increased plasma concentration lowers the SIDm, leading to SID6 acidosis. Hyperlactatemia traditionally has been attributed to dehydration and tissue hypoxia. Therefore, hypoperfusion caused by dehydration could have been responsible for the hyperlactatemia in the goats. However, several studies have challenged this hypothesis because hyperlactatemia can occur in presence of normal tissue oxygenation. It has been proposed that hyperlactatemia in critically ill patients is associated with enhanced aerobic glycolytic flux, which is the most important determinant of lactate production. Increased aerobic glycolytic flux in humans with sepsis and septic shock is enhanced by epinephrine secretion , , and respiratory alkalosis. Additionally, in macrophage‐rich organs including the lungs, intestine, and liver (ie, pneumonia, diarrhea, and hepatitis, respectively), inflammatory cytokines (ie, tumoral necrosis factor and interleukin 1) promote aerobic glycolytic flux for adenosine triphosphate production required for respiratory burst during sepsis. Thiamine (vitamin B1) deficiency negatively impacts oxidative energy production and enhances aerobic glycolytic flux. In ruminants on well‐balanced diets, thiamine is produced by bacteria in the rumen but during periods of anorexia or hyporexia, thiamine deficiency can occur. Thus, it can be hypothesized that in the subset of sick goats included in our study, thiamine deficiency could have contributed, at least in part, to tissue hyperlactatemia. , Propylene glycol (PG) commonly is administered to sick goats, especially those suffering from pregnancy toxemia. It is metabolized to lactate and excreted by the liver or kidneys. Therefore, administration of PG to goats with hepatic or renal insufficiency also could have contributed to the hyperlactatemia.

In our study, changes in TPP also were associated with acid‐base disorders in which hypoproteinemia and hyperproteinemia lead to Atot alkalosis and acidosis, respectively. , The Atot alkalosis in goats with haemonchosis likely resulted from protein‐losing enteropathy, whereas it could have resulted from volume expansion (which occurs during pregnancy) in goats with dystocia. In sick goats, Atot acidosis was diagnosed frequently, likely because of a decrease in plasma water content because of decreased water intake and loss of protein‐poor body fluids (eg, hypersecretory diarrhea). Of the goats that presented to our institution with mastitis, most were acute cases, but in chronic cases it is possible that hyperglobulinemia could have affected the TPP concentration, leading to Atot acidosis.

Contrary to the H‐H approach, the sSID approach offers a calculation to determine the USI and their effect on blood pH and HCO3. In our study, USI acidosis was detected in the majority of goats with pregnancy toxemia likely because of increased concentration of BHB. Four goats included in the “other” diseases group also had USI acidosis, 1 goat had end‐stage liver disease, 2 had polytraumatic lesions associated with dog attack, and 1 had peritonitis. The source and nature of the USI in critically ill patients is unknown, but their concentrations are increased in humans and calves with liver and kidney injuries , and in people with severe trauma. Proposed sources of USI in critically patients include intermediate products from the Kreb's cycle , and exogenous anionic compounds in antimicrobials or electrolytic solutions. Changes in plasma phosphate and protein concentrations affect Atot calculation and, therefore, the USI concentration. , Marked increases in plasma phosphate concentration and production of acute phase inflammatory proteins could contribute, at least in part, to the increase in USI. Finally, USI also could include intracellular anions that are released into the systemic circulation during endotoxemia.

The concentration of USI and values of blood pH on admission were associated with higher hazard ratios of hospital mortality in sick goats. Our models indicated that for every 1 mEq/L increase in USI, measured at admission, the hazard for nonsurvival, at any point in time during hospitalization, increased by 13%. In addition, for each 0.01 pH unit increase measured at admission, the hazard for nonsurvival to hospital discharge decreased by 6.6%. The experimental design of our study prevented us from identifying a true cause and effect relationship between pH and USI and mortality, but acidemia and USI acidosis are important markers of disease severity in human , , , , , , , and veterinary , , , critically ill patients. Of interest, USI was a better predictor for hospital mortality of humans with blunt and penetrating trauma than were pH, HCO3 −, L‐lac−, AG corrected by albumin and standard base excess. Thus, clinicians should consider measured pH and calculated USI in sick goats, but the predictive accuracy of assessing these metrics alone is low with low sensitivity and specificity in identifying non‐survivors. Therefore, it is necessary to assess both USI and pH in light of the battery of information required to establish a treatment plan and a patient's prognosis.

Our study investigated the acid‐base status of a selected group of sick goats using the sSID approach, and the clinicopathological variables associated with the mortality of sick goats. The results likely were biased toward sicker patients because the sample included only sick goats referred to a single teaching hospital. Therefore, our results cannot be directly extrapolated to any other population. However, the cases admitted to our teaching hospital included sick goats with a variety of diseases similar to those admitted to other tertiary referral hospitals. The retrospective design of our study prevented identification of preanalytical effects such as the exact time from sample collection to processing, the type and size of evacuated blood collection tube, and the potential for underfilling of the tube on the measured variables, especially PvCO2, PvO2, and total CO2 concentration. Generally, at our institution blood sample collection is performed using 2 mL plastic tubes containing sodium heparin. The collection is continued until blood stops flowing and aspiration of air into the tube is avoided during the process. The tubes are inverted several times to mix the blood with the anticoagulant, and the collected samples are processed immediately. Although this sampling process could have minimized the effects of preanalytical factors on the measured variables, the results of our study should be interpreted taking this limitation into consideration.

Because of the retrospective nature of our study, data collection was not standardized, and categorization of clinical signs and treatments was not possible. This prevented assessment of the effect of individual treatment variations on the survival analysis. The outcome of the survival analysis could have been affected by the number of euthanized goats included in the nonsurvival group. However, the diagnosis of most euthanized goats carried a poor prognosis even in the presence of unlimited economic resources. A recent study investigating factors predicting mortality in 1400 sick calves with different diseases and diarrhea conducted an analysis of nonsurvival calves by allocating them to the survival group if unlimited financial resources would likely have resulted in their survival. The results of this analysis showed that approximately 80% of the euthanized calves were likely to have died despite the availability of unlimited financial resources. This observation suggests that, in the majority of cases, the clinician's recommendation of patient euthanasia is appropriate and appears to be based on the probability of the animal's survival and welfare rather than economic constraints.

In conclusion, we showed that the sSID approach is a useful methodology to quantify acid‐base disorders in goats and to determine their mechanism of development. We also demonstrated that USI and pH can aid in establishing the prognosis of sick goats. Additional investigations evaluating the dynamic of USI during hospitalization and their association with morbidity and mortality in sick goats are warranted.

CONFLICT OF INTEREST DECLARATION

Authors declare no conflict of interest.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Supplementary Table 1 Admission values of selected clinical and laboratory variables of 143 sick goats.

Click here for additional data file.

Supplementary Table 2 Admission values (median and range) of blood gas, electrolytes, acid‐base and selected laboratory variables of 143 sick goats.

Click here for additional data file.

Supplementary Table 3 Results of the univariable Cox proportional hazard models assessing the association between selected admission clinical and laboratory variables and outcome of 143 sick goats.

Click here for additional data file.