Ruminal acidosis and its definition: A critical review

Ruminal acidosis occurs as a continuum of disorders, stemming from ruminal dysbiosis and disorders of metabolism, of varying severity. The condition has a marked temporal dynamic expression resulting in cases expressing quite different rumen concentrations of VFA, lactic acid, ammonia, and rumen pH over time. Clinical ruminal acidosis is an important condition of cattle and subclinical ruminal acidosis (SRA) is very prevalent in many dairy populations with estimates between 10 to 26% of cows in early lactation. Estimates of the duration of a case suggest the lactational incidence of the condition may be as high as 500 cases per 100 cows in the first 100 d of lactation. Historical confusion about the etiology and pathogenesis of ruminal acidosis led to definitions that are not fit for purpose as acidic ruminal conditions solely characterized by ruminal pH determination at a single point fail to reflect the complexity of the condition. Use of a model, based on integrated ruminal measures including VFA, ammonia, lactic acid, and pH, for evaluating ruminal acidosis is fit for purpose, as indicated by meeting postulates for assessing metabolic disease, but requires a method to simplify application in the field. While it is likely that this model, that we have termed the Bramley Acidosis Model (BAM), will be refined, the critical value in the model is that it demonstrates that ruminal acidosis is much more than ruminal pH. Disease, milk yield and milk composition are more associated with the BAM than rumen pH alone. Two single VFA, propionate and valerate are sensitive and specific for SRA, especially when compared with rumen pH. Even with the use of such a model, astute evaluations of the condition whether in experimental or field circumstances will be aided by ancillary measures that can be used in parallel or in series to enhance diagnosis and interpretation. Sensing methods including rumination detection, behavior, milk analysis, and passive analysis of rumen function have the potential to improve the detection of SRA; however, these may advance more rapidly if SRA is defined more broadly than by ruminal pH alone


INTRODUCTION
Ruminal acidosis has been the topic of numerous studies and substantive review papers (Owens et al., 1998;Nagaraja and Titgemeyer, 2007;Kleen and Cannizzo, 2012;Plaizier et al., 2022;Nasrollahi, 2023).Ruminal acidosis is a very significant disorder of cattle; studies in Wisconsin found a prevalence of 20.1 and 23% of cows that had "subacute acidosis" as defined by rumen pH < 5.5 (Oetzel et al., 1999;Oetzel, 2004); others in Ireland found a prevalence of 11% (O'Grady et al., 2008).A large Australian study found that 10% of dairy cows < 100 d in milk had "ruminal acidosis," as defined by assessment of ruminal VFA, ammonia, lactic acid, and pH measures (Bramley et al., 2008) and a multi-country study found an overall prevalence of 26.1% (Golder et al., 2023a) in dairy cows < 100 d in milk when sampled 2 to 3 h after feeding using the same diagnostic criteria as Bramley et al. (2008).Supplementary Table 1 (Golder and Lean, 2024) provides details of estimates of prevalence and the methods used to define a case.Therefore, it is likely that many cows will experience some level of "ruminal acidosis" during lactation, indeed, some may be affected many times.It can be estimated that if the prevalence of ruminal acidosis is 10% (Bramley et al., 2008) and the duration of a case is 2 d based on data by Golder et al. (2014b), then there would be an incidence of approximately 500 cases over the first 100 d lactation per 100 cows.Understanding and controlling ruminal acidosis is, therefore, critical to ensuring animal well-being and production.
Clinical ruminal acidosis (CRA) is well-characterized; however, the more prevalent condition, "subclinical ruminal acidosis" or "ruminal acidosis" (SRA) often referred to as "subacute ruminal acidosis (SARA)" is Ruminal acidosis and its definition: A critical review H. M. Golder 1,2 and I. J. Lean 1,2 * poorly defined.N. N. Jonsson (Institute of Biodiversity, Animal Health and Comparative Medicine, University of Glasgow, Glasgow, United Kingdom; personal communication) noted that "subacute" is a poorly defined descriptor of the time-course of a disease and is often misinterpreted as either subclinical disease or disease in which clinical signs are mild and that ruminal pH is a poor indicator of disorder.Plaizier et al. (2022) stated "It is also likely important to consider ruminal changes beyond just ruminal pH when studying the effect of rumen metabolism on animal health."Golder et al. (2014c) reported changes in bacterial community composition associated with high levels of concentrate feeding for diets containing pasture, but these differences were closely related to changes in ruminal ammonia, valerate, propionate, and butyrate concentrations.Total VFA concentrations and ruminal pH were less associated with bacterial community composition.Therefore, if we concentrate solely on ruminal pH, we may be oblivious to many other changes in ruminal physiology that may greatly affect the microbial ecosystem and its interaction with the host.Calsamiglia et al. (2012) asked whether SARA was a pH related problem and concluded that it was partially related, but diet had a substantial influence.We concur with the concerns around definition of ruminal acidosis raised by N. N. Jonsson (personal communication) and (Plaizier et al., 2018;Plaizier et al., 2022).
We propose that the lack of precision and clarity of definition for acidotic conditions will lead to suboptimal progress in understanding, researching, preventing, and treating the condition.Non-differential errors in disease classification for dichotomous variables result in weaker associations and drive hypotheses toward the null; however, differential errors in classification of disease result in unpredictable outcomes (Fleiss, 1981;Kelsey et al., 1996).Flegal et al. (1991) demonstrated that with a continuous variable, misclassification can result in differential errors and that the degree and direction of differential misclassification will vary with the exposure distribution, the category definitions, the measurement error distribution, and the exposure-disease relation.For example, Nasrollahi et al. (2017) found that associations between ruminal pH and milk fat and protein percentage were not as strong for indwelling probe evaluations of pH as those conducted in the same cows by rumenocentesis.The fluctuation in ruminal pH in a day (Denwood et al., 2018) and differences in pH within the rumen (Duffield et al., 2004) indicate the likelihood of misclassification using ruminal pH to define this condition.Hence, a more considered approach to the definition and nomenclature used to describe the condition is required.
In this review, we briefly consider the historical descriptions of ruminal acidosis, evaluate the current definitions used to describe acidotic conditions, explore the limitations to use of ruminal pH as a single measure of ruminal acidosis and offer both definitions of ruminal acidosis and diagnostic measures based on strong evidence that can be used in research and the field.A critical part of this review is to evaluate the definitions of ruminal acidosis in the context of medical definitions of disease and to evaluate the evidence for different definitions of ruminal acidosis in the context of Evans postulates as modified by Lean et al. (2009) for metabolic disorders.We evaluate recent findings on the role of substrates in perturbation of the rumen, the ruminal microbiome, the influence of the genome, provide supportive diagnostic measures, and suggest future directions.

HISTORICAL PERSPECTIVE ON RUMINAL ACIDOSIS
Ruminal acidosis is a relatively modern term that supersedes a raft of previous terms and descriptions used for disorders of the rumen induced by an imbalance of organic acids and physically effective fiber (Supplementary Table 2; Golder and Lean, 2024).Acute digestive disturbances in ruminants that may result in death were reported in the 1800s (White and Spooner, 1842).Treatments for "bloat," a disorder that was often confused with ruminal acidosis, date from 60 A.D. from a Roman text that advised "pouring vinegar through the left nostril and putting two ounces of grease in the jaws" (Anonymous, 1958).Textbooks and papers from the mid-20th century described sudden exposure of ruminants to diets high in carbohydrates, usually resulting in death, in terms that included other disorders of ruminant forestomaches (Ryan, 1964).Thus, confusion of clinical signs and nomenclature was evident during this era (Ryan, 1964) and bloat, in particular, was regularly encompassed in the cluster of rumen disorders and potentially confused causal understandings.
Early case reports or anecdotal reviews and, less frequently experiments, described over-eating or "poisoning" with grains, fruits, vegetables, fermented feeds, or other various feeds high in carbohydrate content.For example wheat (Turner andHodgetts, 1949-1959), grapes (Portway, 1957), apples (Merrill, 1952), green fruit (Jensen and Mackey, 1971), silage (Jensen and Mackey, 1971), brewers grain (Owens, 1959), mangolds (Scarisbrick, 1954), fodder beet (Penny, 1954), kiawe beans (Adler, 1949), potatoes (Church, 1976), and green corn (Church, 1976) were all noted as providing a risk of ruminal acidosis.Cereal grains were in general considered the major cause of ruminal acidosis (Church, 1976).Various differentiations of the disorder were described as early as 1842, when acute versus chronic ingestion were mentioned (White and Spooner, 1842).Irwin (1956) categorized symptoms from over-eating of fruit into 3 forms 1) the mild form, 2) the immediate form, and 3) the comatose form.While Portway (1957) described symptoms after excessive grape consumption in categories of peracute, acute, and subacute.Fox (1970) referred to 1) simple ingestion, 2) ingestion with toxemia, 3) and ingestion with impaction and noted that simple ingestion was usually at the individual level rather than herd level.No doubt the spectrum of symptoms and severities that are associated with this disorder has contributed to the confusion in diagnosis and nomenclature.
The breakdown of glucose by rumen microorganisms to first lactic acid and pyruvic acids was demonstrated by Woodman and Evans (1938).By the 1950s associations were established between the accumulation of lactic acid in the rumen, low rumen pH, and clinical signs of "ingestion."Ruminal acidosis was even referred to as D-lactic or lactic acidosis as accumulation of lactic acid which was "produced" in the rumen by Streptococcus bovis and Lactobacillus spp. was considered to be the major contributing factor to rumen acidity and the development of ruminal acidosis in the middle of the 20th century (Turner andHodgetts, 1949-1959;Dunlop and Hammond, 1965).Dain et al. (1955) associated histamine concentrations and a lower rumen pH in the rumen with clinical signs of illness in sheep experimentally over-fed with new wheat or cracked corn in a series of experiments.Ensiling with sulphuric and hydrochloric acids caused severe ruminal acidosis (Brouwer, 1961) and the association between lipopolysaccharide (LPS) from Gram-negative bacteria and ruminal acidosis was described by Mullenax et al. (1966).
By 1970, the condition was more commonly referred to as "acidosis" or "ruminal acidosis" in texts, but use of other synonyms persisted.Dirksen (1970) noted that the majority of papers published between 1950 to 1970 described peracute or severe acute forms of ruminal acidosis in which pathogenic factors were not uniform and suggested that other forms of "ruminal hyperacidity" existed that needed to be distinguished from the acute form.Dirksen (1970) proposed that the clinical designation of "ruminal acidosis" was a collective term for digestive disturbances with non-physiological depression in rumen pH.He stressed that ruminal acidosis was not only caused by the quantity of readily digestible carbohydrates but the proportion of readily digestible carbohydrates to crude fiber and that factors such as water intake, salivation, passage rate of ingesta, absorption, and protozoa may be involved (Dirksen, 1970).Almost 30 years later Owens et al. (1998) similarly stated that the term "acidosis" was collectively used for rumen and intestine digestive disturbances and referred to rumen pH as being one of the benchmarks for defining the severity of ruminal acidosis.Britton and Stock. (1986) regarded acidosis as not one disease, but rather a continuum of degrees of acidosis and later as a continuum of degrees of rumen acidity (Britton and Stock, 1989).But nevertheless characterized it as acute and subacute, for simplicity, to represent both extremes, noting the point at which subacute becomes acute is insignificant and challenging to determine.Britton and Stock. (1986) also noted that the effects of subacute acidosis while more difficult to assess, are just as real as those of acute acidosis, with the most important response being a decrease in feed intake.
The wider adoption of differentiation of SARA from acute or lactic acidosis began in the 1990s with a series of papers by K. Nordlund and E. Garrett that supported the concepts of Britton and Stock. (1986).They used single time point pH cut points based on rumenocentesis samples to define SARA at the herd level; a pH of ≤ 5.5 was considered abnormal (SARA), 5.6-5.8 as marginal (at risk of SARA), and > 5.8 as normal (Nordlund and Garrett, 1994).The diagnosis was focused on sampling a set number of cows from groups of cows with particular demographics and were refined over their series of papers (Nordlund and Garrett, 1994;Nordlund et al., 1995;Garrett et al., 1999) as described in Supplementary Table 3 (Golder and Lean, 2024).When > 30% of samples from any one group had a pH of ≤ 5.5 in conjunction with supporting findings from the history and physical examination of the herd, these were categorized as abnormal (Nordlund and Garrett, 1994).The authors were explicit in stating that pH data cannot be used in isolation for diagnosis and must be used in conjunction with physical examination of the herd.The approach to diagnosis by Nordlund and Garrett (1994) was designed to be practically applied in the field, rather than being definitively accurate.Subsequently researchers failed to adopt these robust criteria and used rumen pH cut points, in isolation from physical examinations, as the "gold standard" for diagnosis of SARA.McArthur and Miltimore (1968) were among the first reported use of constant monitoring with an indwelling pH probe suspended in place in the ruminal ventral sac.In experimental facilities with access to cannulated cattle, researchers measured rumen pH at regular intervals, commonly 3 h intervals, across a 24-h period to define SARA (Krehbiel et al., 1995;Beauchemin et al., 2001).To further account for rumen pH fluctuations across a 24-h period continuous measurement of rumen pH with an indwelling pH probe in the ventral sac placed via a fistula was used in experimental facilities to define SARA.Cooper et al. (1997) reported the average rumen pH, daily pH change, daily variance of rumen pH, and area of pH (ruminal pH units below 5.6 by min).Other early adopters used indwelling probes that took a rumen pH measurement every second and reported average pH, Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW time below an average pH of 5.6 and pH 6.0 in min/d, and area (time × pH) below pH 5.6 and pH 6.0 reported in min × pH/d in 24-h periods (Cumby et al., 2001;Krajcarski-Hunt et al., 2002;Gozho et al., 2005) or similar metrics.Beauchemin et al. (2003), Gozho et al. (2006), andGozho et al. (2007) concluded that the findings of Gozho et al. (2005) and Cooper et al. (1999) showed rumen pH values < 5.6 for ≥ 3 h indicated SARA which became the new standard for SARA definition.Dado and Allen (1993) compared rumen pH from an indwelling pH probe with manual pH measurements, finding a total error between methods of 0.17 pH units with 41% from mean bias and 54% from random disturbance.The correlation was 0.85.The various cut points and areas under the curve (AUC) for pH used in studies are provided in Supplementary Table 4 (Golder and Lean, 2024).Neither McArthur and Miltimore (1968) or Dado and Allen (1993) used their devices to define SARA but provided valuable precursor work to the development of the current devices used today for the study of rumen kinetics and classification of ruminal acidosis.Villot et al. (2018), for example used relative daily pH measures calculated from continuous indwelling boluses to determine SARA and non-SARA cattle and associated these classifications with milk and rumen measures, and the ruminal bacterial species present.
The limitations of using rumen pH as a sole parameter for diagnosis of ruminal acidosis has become more evident, and there is growing uncertainty on its interpretation as a diagnostic measure.The variability in pH within a rumen and the limitations to measuring pH with probe accuracy and drift in pH measurement with indwelling devices increases measurement error.Specifically having rigid pH cut points for classification of ruminal acidosis has led to alternate, multivariable approaches to its classification (Brown et al., 2000;Bramley et al., 2008) but these are yet to be widely adopted.With greater use of integrated metrics, advances in nanotechnologies and sensors, we expect new models for rapid and accurate classification to become readily available for commercial herds.
While ruminal acidosis continues to be known by a range of synonyms, diagnosis has moved from that of individual cows with clinical ruminal acidosis, to single time point herd level diagnosis of non-clinical forms, to continuous monitoring of ruminal pH in experimental facilities, to most recently, diagnosis by multifactorial measures.It is important to note that continuous monitoring using fistulae or indwelling probes with wireless transmission of data provide robust estimates of one aspect of ruminal change in small numbers of cattle, while point sampling methods such as rumenocentesis and stomach tube samples are potentially valuable to study larger populations.It is probable that as the potential to monitor individual feed intake and rumination increases, decreased DMI and or rumination may be useful monitoring tools.A lack of data from the field, of uniform definition, and clinical diagnosis was noted in a review by Kleen et al. (2003); however, 2 decades later these issues remain and emphasize the need for a critical and quantitative review of ruminal acidosis.

CHALLENGES OF DEFINITION AND DEFINITIONS
Definitions of ruminal acidosis have been derived through associations with an observed change in rumen pH and feeding carbohydrate rich diets and feeds, rather than necessarily describing the patho-physiology of the condition.Acidotic conditions of the rumen can be considered a continuum of conditions of varying severity (Britton and Stock., 1986;Lean et al., 2014) and with marked temporal dynamic expression in DMI, rumen VFA, lactic acid, and ammonia concentrations (Figure 1).This continuum of varying severity reflects the challenge of safely sequestering hydrogen that accumulates from carbohydrate fermentation.Safe pools to reduce or delay the availability of rapidly fermentable substrates, hydrogen, and hydrogen protons include starch engulfment by protozoa, bacterial glycogen formation, growth of bacteria, methane, and weak organic acids (Figure 2).The role of uptake and removal by active and passive transport mechanisms of VFA and lactic acid to remove protons from the rumen is important and well described (Aschenbach et al., 2011).We refer to VFA synonymously with short chain fatty acids (SCFA).Less safe pools include lactic acid, an acid that is 10 times stronger than the VFA.Decreasing the hydrogen supply by utilizing more fiber and feeds that are fermented slower, thereby enhancing rumination can reduce the risk of ruminal acidosis.It is important to recognize that the effects, and possibly even pathogenesis of ruminal acidosis may not be solely ruminal and other parts of the gastro-intestinal tract may play a role.However, the latter is not a focus for this paper.

Clinical Ruminal Acidosis (CRA)
A review of the definition of medical conditions shows that acute disorders have a rapid onset, are clinically evident, and are of short duration (King, 2013;Studdert et al., 2020).Clinical ruminal acidosis (CRA) or lactic acidosis has a rapid onset, marked clinical signs, and short duration; however, chronic sequalae can result.
Clinical ruminal acidosis is caused by the sudden access of cattle to rapidly fermentable carbohydrates (RF-CHO), changed processing of the same RFCHO, or even novel and abrupt introduction of moderately fermentable carbohydrates and the clinical diagnosis is usually ob-Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW vious from history and clinical signs.Glucose has been used to create CRA (Nagaraja et al., 1981) and fructose appears to have greater potential to cause SRA and CRA than starches (Golder et al., 2012b;Golder et al., 2014b).Clinical ruminal acidosis is a serious condition of cattle characterized by death, possible in 8 to 10 h (Underwood, 1992), dehydration, ruminal distension, diarrhea (often with grain in the feces and a sickly, sweet smell), abdominal pain, tachycardia, tachypnea, staggering, recumbency, coma, a marked decline in milk yield, and sequalae including ruminitis, liver abscess, pulmonary infections, epistaxis, and poor production, either weight gain or milk, that arises subsequent to the ingestion of large amounts of RFCHO.The rumen fluid can be milky white often containing grain on examination.
Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW Figure 1.Mean (±SEM) for (A) distance to acidosis centroid groups by supplement feeding amount and sample time, (B) rumen pH by supplement feeding amount and sample time (C) concentration of propionate by supplement feeding amount and sample time, and (D) concentration of valerate by supplement feeding amount and sample time.All cattle were offered 15 kg of DM/cow per day with a target pasture intake of 8 kg DM/cow per day plus 1 of 5 supplement amounts (ryegrass silage and crushed wheat at a ratio of 27 to 73, respectively).The ryegrass silage was fed under an electric wire on the pasture and the crushed wheat twice daily at milking.The units were kg of DM of total supplement/cow per day.n = 2 cows/supplement amount.Sample times were collected approximately 2.4 h apart over a 24-h period.Sample time 1 was approximately 0820 h and milking/feeding was at 0700 and 1500 h (black arrows).Sample time 8 was excluded.A distance to the center of the acidosis cluster (eigenvalue) of 0 corresponds to a healthy, non-acidotic rumen sample and 1.0 represents an acidotic sample.Data from Golder et al. (2014c).
Clinical ruminal acidosis is defined by the generation of significant amounts of lactic acid in the rumen; acute ruminal acidosis is present when rumen pH is < 5.0, there is > 50 mM lactic acid and ruminal VFA are less than 100 mM (Nagaraja and Titgemeyer, 2007;Golder et al., 2014a;Golder et al., 2014b).The diagnosis should not be based on lactic acid concentrations alone, rather be interpreted in conjunction with clinical signs and diet.There is a general consistency of definition and understanding of CRA which is the most extreme expression of the continuum of disorders related to ingestion of rapidly degrading carbohydrates in the rumen and has high morbidity and mortality if untreated.

Subclinical Ruminal Acidosis (SRA)
The term "subacute" has frequently been applied to ruminal conditions in which there is a decrease in ruminal pH as determined from a single sample or as time spent below an arbitrary ruminal pH measured using indwelling probes such as rumen boluses, or with pH probes inserted through fistulae.Dictionary definitions whether derived from human medicine (King, 2013), veterinary medicine (Studdert et al., 2020), andEnglish (Oxford Dictionary, 1971) or American (Dictionary, 2014) language note the vague nature of the term which is applied to a disease condition that is neither acute, nor chronic.The terms "acute," "subacute," and "chronic" are predominately defined by units of time to classify the severity of disease or pain in the medical profession based on causation and prognosis and imply clinical significance and, therefore, treatment (King, 2013).The term "subacute" evolved as a descriptor for the intermediate state between acute and chronic pain present for > 14 d and between 6 wk and 3 mo (Hippocrates.Of the epidemics written 400 B.C.E translated by Francis Adams.; Van Tulder et al., 1997).It is often regarded as a subset of acute (King, 2013), consistent with its definition in veterinary science; "somewhat acute; between acute and chronic" (Studdert et al., 2020).There is no evidence that the conditions associated with a moderate decrease in ruminal pH create immediate pain, and pain in animals is poorly quantified.In-dwelling probe studies show that many asymptomatic cattle have ruminal pH that lie below commonly described cut-points for ruminal pH used to define "subacute acidosis" (Denwood et al., 2018).Further, many individual cow determinations in large populationbased studies (Bramley et al., 2008;O'Grady et al., 2008; Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW Figure 2. Indicates major pathways influencing ruminal acidosis.Rates of hydrolysis are indicated for sugars, starches, and other rapidly fermentable carbohydrates (RFCHO).Safer pools for starch and hydrogen ion sequestration are indicated by green text and include starch ingested by protozoa, microbial protein, microbial glycogen, valerate, propionate, and hydrogen gas.Removal of volatile fatty acids (VFA) that include acetate, butyrate, propionate, valerate, and caproate from the rumen through the ruminal wall result in the influx of bicarbonate ion which is important to maintenance of rumen stability.Golder et al., 2023a) had ruminal pH which were well below thresholds considered "subacute" without apparent clinical signs of illness or pain.
In contrast with "subacute," the term "subclinical" is defined as a disease before clinical symptoms are first noted; without clinical manifestations; said to be of the early stages or a mild form of a disease, e.g., subclinical disease, infection, parasitism, or when a disease is detectable by clinicopathological tests but not by a clinical examination (Studdert et al., 2020).Subclinical conditions such as mastitis, hypocalcemia, and ketosis are routinely reported and are associated with increased risks of other disorders and milk production loss.It is also noteworthy that subclinical conditions are associated with indicator variables and modest associations with other disease and milk production outcomes.For example, increased milk SCC is reflected in milk production loss (Raubertas and Shook, 1982), subclinical hypocalcemia with increased risk of placental/ retention and metritis (Curtis et al., 1983;Rodríguez et al., 2017), and ketonemia with reduced risk of pregnancy (Walsh et al., 2007).While some of these associations among conditions and outcomes are substantial (Rodríguez et al., 2017) others are more subtle (Raubertas and Shook, 1982).The lack of strength of association may well reflect the impacts of misclassification of cases, especially in cross-sectional studies.This potential is suggested from changes in VFA absorption and concentrations following carbohydrate challenges.Schwaiger et al. (2013) identified that 2 d after a challenge with a ruminal infusion with crushed barley at 10% of BW there are marked decreases in ruminal absorption of propionate, acetate, and butyrate and Golder et al. (2014a) found extremely low VFA concentrations on d1 after CRA.The concentrations of VFA increased after 2 d (Golder et al., 2014a) suggesting that misclassification of this condition in observational studies is likely and estimates of effect may, consequently, be conservative.We conclude that the term "subacute" is not appropriate to describe the condition of subclinical ruminal acidosis and suggest that the term "subclinical ruminal acidosis" (SRA) should be used while acknowledging that the etiopathogenesis of this disorder is complex, and no definition will be perfect.
In contrast with CRA, SRA is characterized by low lactic acid concentrations in rumen fluid, but high concentrations of VFA, particularly propionate (>30 mM) and valerate (>2.4 mM), low ammonia concentrations (<3 mM) and a moderately low pH (Morgante et al., 2007;Bramley et al., 2008;Golder et al., 2023a).The following clinical signs and conditions have been associated with SRA; increased risk of lameness (Bramley et al., 2013), cyclic variation in feed intake (DeVries et al., 2014;Golder et al., 2014b), body weight loss (Kleen et al., 2013) and increased risk of low milk fat percentage (Bramley et al., 2008;O'Grady et al., 2008; Table 1).The sequalae including ruminitis, liver abscess, pulmonary infections, abomasal displacement, epistaxis, and poor production are possible, but evidence that these are associated with SRA as well as CRA is required.
The following factors are likely to influence the expression of both CRA and SRA i) production of toxic substances and clearance of these from the rumen (Ametaj et al., 2010;Zhang et al., 2017).Zhang et al. (2017) found increases in concentrations of amino acids, bacterial degradation products including amines, and sugars with increased amount or proportion of concentrates fed.The generation of toxins and clearance of toxins will be influenced by 1) ruminal populations of microorganisms; 2) compromised epithelia, through chemical action within the rumen and, consistently with the BAM model, high concentrations of VFA and low rumen pH (Steele et al., 2011;Meissner et al., 2017) and conditions such as pestivirus, that damage epithelial integrity and reduce the ability to prevent toxins from being absorbed, 3) rate of passage and differential clearance of substrates and metabolites and exposure of different parts of the gastrointestinal tract to these, and 4) inflammatory responses including tumor necrosis factor, interleukin 1β, LPS binding protein (LPB), and interleukin 6 (Emmanuel et al., 2008;Zhao et al., 2018;Aschenbach et al., 2019;Albornoz et al., 2020).Stefanska et al. (2018) identified gene upregulation in white cell immune responses consistent with increased exposure to LPS.These functions may be influenced by mammalian genetics and understanding the interactions of these with the ruminal metabolome and metataxome is important (Golder et al., 2018;Golder et al., 2023c).Consequently, the series of changes caused by the increase in RFCHO extends well beyond a decrease in ruminal pH and includes changes in numerous metabolic pathways involved in ruminal acidosis including the generation of potentially toxic metabolites (Ametaj et al., 2010;Zhang et al., 2017).
There is considerable speculation about agents that might cause some of the clinical signs of CRA and conditions associated with SRA.Lean et al. (2013b) summarized some of the evidence supporting potential roles for histamine, LPS, and lactic acid to cause laminitis (Table 2).Associations between LPS and acute ruminal acidosis were first documented by Mullenax et al. (1966) and Nagaraja et al. (1978).Lysis of Gram-negative bacteria increases during ruminal acidosis as the rumen environment becomes less optimal for their growth and survival (Monteiro and Faciola, 2020).Of the 3 components of LPS, it is the lipid-A portion that primarily leads to toxicity once in portal circulation by attaching to endothelial cell receptors, mainly Toll-like receptor 4, once in the liver which initiates immune responses (Mani et al., 2012).The composition of lipid-A causes the toxicity of LPS varying with the number of acyl groups, which ranges between 4 and 6, with 6 groups being more bioactive (Netea et al., 2002).The potential toxicity of each Gram-negative bacterium is unknown but varies (Monteiro and Faciola, 2020).
The LPS is believed to be translocated into the bloodstream if the rumen epithelium is damaged from the accumulation of acids and subsequent increase in osmotic pressure that promotes paracellular translocation (Owens et al., 1998;Plaizier et al., 2012;Aschenbach et al., 2019).Second, LPS can be transported transcellularly, mediated through toll-like receptor 4 (Mani et al., 2012).Some authors report changes in endotoxin units (EU) in both the rumen and blood (Khafipour et al., 2009b;Zhao et al., 2018), where others reported changes in the rumen but not in the blood (Gozho et al., 2007).A metaanalysis found that increases in ruminal endotoxin were linear when cattle were fed above 44.1% concentrate or below 39.2% NDF and were associated with increased plasma haptoglobin, and serum amyloid A levels (Zebeli et al., 2012).Similar depressions in ruminal pH for cattle challenge fed with alfalfa pellet and ground alfalfa to those observed in cattle challenged with high amounts of concentrate suggest that ruminal pH depressions and increased ruminal LPS concentrations alone do not cause an acute phase response (Plaizier et al., 2012).
Histamine is a vasoactive amine produced by the decarboxylation of histidine that is generated in the rumen after feeding and can accumulate during acidotic conditions (Nilsson, 1963;Ahrens, 1967;Golder et al., 2014a; Tables 1 and 2); however, its involvement in the pathogenesis of ruminal acidosis and its possible association with laminitis remains unclear.While accumulation of histamine is not directly responsible for epithelial damage in the rumen (Ahrens, 1967), it delays epithelial regeneration (Aschenbach et al., 1998).Absorption of histamine can occur across an intact rumen epithelium; however, a combination of low epithelial permeability and inactivation by catabolism of > 90% result in low net absorption of histamine (Aschenbach and Gabel, 2000), a result that supported earlier studies (Kay and Sjaastad, 1974).Epithelial damage appears to increase the net absorption of histamine (Aschenbach and Gabel, 2000); however, once absorbed, histamine is rapidly metabolized to inactive forms by either methylation or oxidation (Goth, 1974).Histamine can also be produced endogenously by activation of mast cells resulting in release into circulation.Krogstad and Bradford (2023) reviewed the changes in inflammatory biomarkers to increased starch intake that are triggered by exposure to challenge and noted the variability of LPS and acute phase protein responses to starch challenge.Results provided in Table 2 support the Krogstad and Bradford (2023) findings and the need for a greater understanding of the pathogenesis of SRA.Given the large number of potential toxins, often produced simultaneously in the rumen (Ametaj et al., 2010), a singular focus on any particular toxin may not be appropriate.Given the known agents capable of causing inflammation and clinical signs, and that less well-known metabolites may be involved in clinical signs of ruminal acidosis, it is unsurprising that rumen pH per se is largely unrelated to the clinical signs of ruminal acidosis.Bramley et al. (2008) collected diet, health, production, and rumen fluid data from 800 cows from 100 randomly selected commercial dairy herds across Southern Australia to establish a robust strategy for the diagnosis of ruminal acidosis.The intent was to use the observations made to determine whether an uninformed approach to categorization would be superior to using rumen pH cut points to define cases of ruminal acidosis.The data were used to create a model by k-means cluster and discriminant analysis that classified cattle based on their rumen pH, individual VFA, ammonia, and D-lactate concentrations into one of 3 categories (1) acidotic, (2) suboptimal rumen function, or (3) normal (Bramley et al., 2008;Golder et al., 2012a).The prevalence of individual cattle in each of those 3 categories was then used to categorize the 100 herds into 3 categories of risk.Hence both individual and herd-level risks for ruminal acidosis were defined.Although the diet, health, and production data were not included as inputs into the model, their associations were used to support and define the characteristics of the categories (Bramley et al., 2008;Bramley et al., 2013).The characteristics of each category for acidosis are described in Bramley et al. (2008).Bramley et al. (2008) conducted their study on a wide range of herds that fed only pasture, through to different levels of grain and supplement feeding including TMR herds.Herd was not a significant factor in the study in the prediction of ruminal acidosis.Similarly, herd was not significant in Golder et al. (2023a) in a study that included cows from 3 different countries and 32 herds with a predominance of TMR fed herds.Findings indicate that changes in ruminal fermentative responses to dietary substrates are a function of bovine metabolism rather than being profoundly influenced by production system and indicates robust applicability of the model.Tightly controlled challenge studies using 1.2% of BW fed as grain on a DM-basis and repeated measurements over several hours, showed that propionate, ammonia, and valerate concentrations were the most sensitive indicators of the potential for different grains to cause ruminal acidosis (Lean et al., 2013a), and that the Bramley  2008) on a separate occasion (unpublished data).In addition, a randomized clinical study identified a profound difference in risk of ruminal acidosis, as defined by the BAM with increased intake of concentrates (Golder et al., 2014c); therefore, meeting several of Evans' postulates as modified by Lean et al. (2009).

Bramley Acidosis Model
The BAM can be used to produce an acidosis eigenvalue (probably of distance from a known case of ruminal acidosis with a value between 0 and 1; 1 being acidotic) and identify the acidosis risk category (3 possible categories) for one or more cows at the time of rumen sampling that are not in the original data set.This can be done by appending the individual VFA, ammonia, D-lactate, and pH values (either from rumenocentesis or stomach tube) for these cows to those of the existing data set of Bramley et al. (2008) that was used to develop the k-means cluster analysis group allocation.The data set and details on how to utilize it are provided on FigShare (Golder and Lean, 2024).Our data suggest that the best time to sample is 2 to 3 h after feeding; Figure 1 indicates the impact of sampling time.Bramley et al. (2008) described the characteristics of their 3 acidosis categories as corresponding to acidotic, suboptimal rumen function, or normal, but when other data sets are added these terms may not best represent the data.For example, Golder et al. (2023a) used the terms high, medium, and low risk, where the characteristics of category 2, suboptimal rumen function group, from Bramley et al. (2008) better aligned with the characteristics of a low-risk group.
The lactic acid concentrations in the Bramley et al. (2008) data set were mostly low, consequently the model identifies SRA.The model is not robust when lactic acid concentrations are high (we suspect possibly > 30 mM).The data from Golder et al. (2014b) were successfully categorized by the model except for responses when heifers were challenged with 1% BW of milled wheat and 0.2% BW of fructose.The latter responses were characterized by mean ruminal concentrations of total lactic acid per treatment group between 10 and 45 mM over a period up to 215 min after challenge and were considered to be "acute ruminal acidosis" or CRA (Golder et al., 2014b).One control heifer from this challenge study by Golder et al. (2014b) had a total lactic acid concentration of > 100 mM on challenge day that only reduced to 70 mM on the subsequent morning with clinical findings described in a case report (Golder et al., 2014a)  Appears to be probable.

The Application of Evans Postulates as Modified to Subclinical Ruminal Acidosis (SRA)
Evans ( 1976) provided a detailed commentary on existing methods for defining the causation of infectious disease and developed a set of widely acknowledged postulates to be applied to defining causation of infectious disease.Some of these definitions were modified by Lean et al. (2009) to define metabolic disorders.The definitions are tested for the condition of SRA as categorized by using either rumen pH alone or by using a model that evaluates the VFA, lactic acid, ammonia, and pH in rumen fluid (Bramley et al., 2008).
Factors that can cause ruminal acidosis include a high or rapidly increasing dietary starch intake (Li et al., 2012b), a short previous exposure to diets high in carbohydrate (Schwaiger et al., 2013), reduced particle size or greater ruminal availability of starch (Schwaiger et al., 2013;Lean et al., 2014), increased sugar intake (Nagaraja et al., 1981;Golder et al., 2012b), changing the forage fed and fiber content (Plaizier et al., 2008;Khafipour et al., 2009a), lack of available protein or nitrogen (Golder et al., 2014c), and changing the particle size of the feed (Zebeli et al., 2012).These attributes provide diets that increase ruminal availability to rapidly fermentable carbohydrates or reduce the fiber content of the diet or effectiveness of dietary fiber to stimulate saliva production and ruminal buffering.We propose that these dietary changes represent the conditions that characterize "exposure" to the continuum of acidotic conditions that may be categorized as SRA and CRA and can be tested using the following postulates.

Postulate 1) the definition of disease should be consistent with current understandings of the biochemical basis of the disorder or reflect modifications thereof
Figure 2 summarizes biochemical pathways which are critically involved in carbohydrate and protein metabolism that influence the risk of ruminal acidosis.Some aspects of quantitative and temporal contribution to the pathogenesis such as ingestion of starch by protozoa, increases in microbial glycogen, and bacterial production of exopolysaccharides (Hackmann and Firkins, 2015) are yet to be determined, but have been evaluated in vitro (Hall, 2011;Hackmann and Firkins, 2015) and will vary across cattle populations and environments.
The biochemistry of rumen pH, the most widely used diagnostic tool, is complex and there are compelling reasons for ruminal pH to be controlled; hydrogen must be sequestered into safe pools as strong acids damage tissues and microbiota.Cattle with the same pH can have H + ions sequestered in different molecules with different pKa and H + ions may be sequestered in "safe" and "unsafe" sinks within the rumen with completely different physiological outcomes.Lactic acid concentration, for example, does not always correlate with rumen pH (r = −0.14;Britton and Stock, 1989), nor does it strongly reflect the most common form of ruminal acidosis in dairy cows, SRA.Table 1 summarizes biochemical changes in studies that used rumen pH cut points or AUC to categorize ruminal acidosis risk.Further, the rumen is dynamic and not homogenous and any measure of pH whether continuous and indwelling, or static and singular has limitations.Evacuation of the rumen provides a homogenized measure, but not one that is physiologically relevant.Similar limitations exist when considering the biochemical changes that occur in ruminal acidosis, as rumen function varies within the rumen mat, liquid phase, and near the rumen wall and papillae (Penner, 2014).Table 3 shows the differences and correlations between different measures of rumen pH.Bramley et al. (2008) tested both rumenocentesis and stomach tube ruminal pH and found that the rumen pH measures were not highly predictive for the group of cows that were more prevalent in herds where dietary NFC were higher and NDF lower.Ruminal pH was lower in the acidotic group of cows than others but was the least strongly related variable to the category (Bramley et al., 2008); a finding that is consistent with the challenges in obtaining representative pH determinations from a highly heterogenous environment (Table 3).Rumen pH will decline with exposure to the risk factors for ruminal acidosis as identified in numerous studies but is less descriptive of the biochemical changes evident in Figure 2 and Table 1.
Evaluating ruminal acidosis using more measures than rumen pH provides greater insight to the biochemistry of ruminal acidosis.Tables 1 and 4 indicate that cattle with "ruminal acidosis" as defined by Bramley et al. (2008) have characteristics that would be expected with current understandings of biochemical responses to increased intake of highly fermentable carbohydrates and other exposure factors.Concentrations of ruminal VFA reflect rates of production and irreversible loss from interconversion, outflow, and absorption.Xue et al. (2018) explored relationships among ruminal bacteria with the VFA produced, subsequently with milk and milk component production for more than 300 cows on the same diet demonstrating the variability of the microbiome among cows and the central role of VFA production to milk and milk protein production.Subsequently, Xue et al. (2020) explore in detail the multi-omic interdependencies of substrate, bacteria, and host indicating the robustness of the rumen.Normal concentrations of ruminal total VFA are between 70 to 130 mM with the ratio of acetate to propionate to butyrate usually approximately 70: 20: 10, respectively, for cattle fed forage diets (France and Dijkstra, 2005).
In general, during SRA or with increased fermentable CHO feeding, total VFA concentrations increase, with in-Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW creases in propionate concentrations to approximately 35 to 45% of the total pool at the expense of acetate (Ørskov, 1986;France and Dijkstra, 2005).Propionate is one of the end products of both the succinate and acrylate pathways (Figure 2).The acrylate pathway converts lactic acid (pKa 3.85), a less safe proton sink, into propionate (pKa 4.87), which is a safer proton sink and lowering the risk of ruminal damage from lactic acid, noting that both acids will be largely dissociated at pH 5.8.Thus, high concentrations of propionate relative to normal can be perceived as being protective or detrimental to the rumen depending on the source of substrates, buffering capacity of the rumen, and fermentation pathway.Receiver operator characteristic curves produced from the data sets of Bramley et al. (2008) and Golder et al. (2023a) (Table 4; Figure 3) showed that propionate was both a sensitive (0.95) and specific (0.81) indicator of SRA.The AUC (0.88) was of very good diagnostic value.
During ruminal acidosis both increases (Khafipour et al., 2009b) and decreases in butyrate concentrations have been reported (Kennelly et al., 1999), perhaps reflecting the source of substrate.Table 4 and Figure 3 shows that butyrate was neither sensitive (0.65) nor specific (0.64) for SRA and the AUC (0.57) is similar to chance (Table 4).
High concentrations of valerate have been associated with ruminal acidosis (Enemark et al., 2004;Morgante et al., 2007;Bramley et al., 2008;Nasrollahi et al., 2017; Table 1) and may indicate that valerate is acting as a safe hydrogen sink for the removal of lactic acid.Valerate is usually present in relatively low rumen concentrations but when lactate is available, valerate can be formed from this substrate (Stewart et al., 1997).Thus, high concentrations (>2.4 mM), are an indicator of rumen perturbation and may act to protect the rumen from damage, albeit in a limited context.The ROC results (Table 4; Figure 3) produced from the data sets of Bramley et al. (2008) and Golder et al. (2023a) showed that valerate was both a sensitive (0.95) and specific (0.81) indicator of SRA and the AUC (0.89) was of excellent diagnostic value.Caproate concentrations are often analyzed as are iso-butyrate and iso-valerate and increased caproate concentrations are associated with cattle with SRA (Golder et al., 2023a).Caproate can be produced from glucose fermentation (Marounek et al., 1989) or 3 moles of lactic acid (Annison and Lewis, 1959).Caproate has moderate sensitivity (0.71), specificity (0.74), and AUC (0.73) (Table 4; Figure 3).Bramley et al. (2008) found that low ruminal ammonia concentrations were associated with SRA.Similarly, Golder et al. (2023a) found low concentrations of ammonia in cows with SRA but, conversely high concentrations occurred in a heifer with CRA (Golder et al., 2014a), likely reflecting the difference in pathogenesis between SRA and CRA.Microbial protein is a significant safe sink for hydrogen in the rumen and the availability of peptides and ammonia will allow rapid growth of the bacteria from the phylum Firmicutes (Figure 2).Golder et al. (2023b) identified 5 phyla and 9 families associated with increased risk of ruminal acidosis and these were associated primarily with sugar and protein intake.Lower ammonia concentrations of SRA cows provide associations with microbial protein synthesis and energy spilling (Nocek and Russell, 1988;Hackmann and Firkins, 2015).Table 4 and Figure 3 show that ammonia concentrations of < 44 mM were not highly sensitive (0.68) but were specific (0.79) and moderately useful as a diagnostic test for SRA (AUC = 0.74).The mean ammonia concentrations were 74.6 mM (SD 54.9) from the combined (Bramley et al., 2008) and Golder et al. (2023a) data set.Thus, increased production of microbial protein both sequesters hydrogen and reduces production of VFA.
Lactic acid concentrations are often in flux as lactic acid in the rumen can be absorbed into the bloodstream, accumulate in the rumen, pass out of the rumen with ingesta, or be fermented, primarily, to propionate, butyrate, caproate, isobutyrate, and valerate (Annison and Lewis, 1959;Huntington, 1988;Stewart et al., 1997).Lactic acid is generated rapidly after feed consumption and can be converted into those end products when the rumen is able to control this influx of acid.The rumen epithelium doesn't have a known specific active transport system for lactic acid resulting in low rates of absorption (Williams and Mackenzie, 1965;Aschenbach et al., 2011).If the rumen is unable to remove the amount of lactic acid produced it can accumulate over a period of hours until reaching a peak and subsequently declining.Thus, time of rumen fluid sampling relative to feeding, and the cow's ability to control lactic acid accumulation influences observed concentrations in the rumen.
It has been suggested that accumulation of ruminal lactic acid only occurs in cattle with "acute ruminal acidosis," as opposed to those with "SARA" (Garrett, 1996;Oetzel et al., 1999) but there is no concensus on this.Normal ruminal total lactic concentrations have been defined as up to 5 mM (Owens et al., 1998;Nagaraja and Titgemeyer, 2007), < 10 mM (Harmon et al., 1985;Burrin and Britton, 1986;Goad et al., 1998), or up to 50 mM (Horn et al., 1979).Owens et al. (1998)  were observed when sugars rather than starches were fed (Harmon et al., 1985;Heldt et al., 1999;Golder et al., 2012b).Lactic acid was not a useful measure for diagnosing SRA (sensitivity 0.65; specificity 0.61 and AUC 0.63; Table 4), or when cattle are adapted to high-grain diets (Huntington, 1988); however, it has value as part of a diagnostic model using other measures (Bramley et al., 2008).
This postulate that understandings of the biochemistry of ruminal acidosis are consistent with definitions of the condition is strongly supported by evidence from studies using the BAM and supported by studies using rumen pH cut points or AUC.
Postulate 2) the proportion of individuals with the disease should be significantly higher in those exposed to the supposed cause than in those who are not There are few studies, that is case-control studies, which provide the evidence to support postulate 2 for SRA and none using an integrated measures approach such as the BAM.Guegan et al. (2015) sampled 144 cows from 12 herds with suspected ruminal acidosis and found significant negative correlations between ruminal pH and rumen fill, ruminal pH and herd milk yield, and ruminal pH and quantity of concentrate per day, providing some support for the postulate.While Enemark et al. (2004) similarly selected herds suspected of ruminal acidosis, only 6 herds were investigated and no detailed diet information was available.It is possible that continuous monitoring of rumen pH with indwelling boluses may provide greater diagnostic value for this postulate; however, Denwood et al. (2018) in a detailed evaluation of the use of these concluded that "future efforts to describe continuously monitored pH data should be based on deviations from an expected predictable rhythm rather than observed measurements below some arbitrarily defined pH threshold."Retrospective case-control studies in populations of herds could provide evidence to support this postulate.
Postulate 3) exposure to a supposed cause should be present more commonly in those with than in those without the disease, when all other risk factors are held constant Bramley et al. (2008) found strong associations with a high herd prevalence of ruminal acidosis with low  NDF, low peNDF, and greater NFC, estimated ME, and starch in diets for the herds.Cows categorized as acidotic also had lower milk fat percentage, similar milk fat yield, and milk fat: protein ratio than those not categorized as acidotic.Consequently, the proportion of individuals with the disease was significantly higher in those exposed to the supposed cause.Herds with higher NFC (>40%) and lower NDF (<31%) and high prevalence of acidotic cows also had a high prevalence (>40%) of lameness compared with the other herds (Bramley et al., 2013), and high prevalence of a milk fat to milk protein ratio of < 1.02 (Bramley et al., 2008).While in a cross-sectional study, one cannot attribute causality, especially for lameness, to the higher prevalence of acidotic cows in the herd, the low milk fat to protein ratio of the tested cows suggests that those cows in exposed herds were at greater risk of ruminal acidosis.Similarly, the multi-country, multiherd study with 32 herds and more than 300 cows found lower milk fat percentage and milk fat: protein ratio for acidotic cows but no difference in milk fat yield (Golder et al., 2023a).O'Grady et al. ( 2008) used the methods of Nordlund and Garrett (1994) to categorize cows and herds as acidotic or not.Milk fat to protein ratio was 1.01, 1.11, and 1.17 in herds categorized as acidotic, high risk, and low risk, respectively, based on prevalence of cows with pH ≤ 5.5; however, there was a very low study power to determine differences among categories (O'Grady et al., 2008).The categories used to define herds and the nutritional evaluation of their diets were not associated nor consistent with increased exposure to rapidly fermentable CHO nor decreased availability of fiber (O'Grady et al., 2008).While study power was low, differences in available substrates, specifically total concentrate fed, pasture CP, ADF, NDF, and water-soluble carbohydrate contents were not consistent with the differences of prevalence of categories within herds.When the data from O' Grady et al. (2008) were tested using the BAM only 50% of the cows below the cut point of rumen pH ≤ 5.5 were considered acidotic.Morgante et al. (2007) also categorized cows using a rumen pH of < 5.5 and found a high prevalence of cases in 3 of 12 TMR-fed   Italian herds.An association was found with higher ruminal concentrations of propionate in these herds and a tendency for higher n-valerate concentrations.There was no evidence for differences in diet among herds with low or a high prevalence of cows with pH < 5.5.Similarly, Kitkas et al. (2013) used a cut point of pH < 5.5 to categorize cows and herds and found no association with feed composition based on NDF, ADF, NFC, or forage to concentrate ratio but found particle size length was associated.Plaizier et al. (2018) highlight many studies that estimated the prevalence of low rumen pH, but cows with low pH did not have significantly different clinical outcomes to other cows, apart from low BCS (Kleen et al., 2013).Kleen et al. (2013) found a tendency toward association between pH ≤ 5.5 and lameness and a lower milk fat to protein ratio for cows with pH ≤ 5.5.The lack of association between diet composition and ruminal acidosis for the studies using pH < 5.5 as a cut point for categorization, contrasts with Bramley et al. (2008).It is notable that the Bramley et al. ( 2008) study had 100 herds and 800 cows with a wide range of diets from pasture to TMR providing a greater study power than pH studies cited for Postulate 3. Postulate 3 is met for the BAM, but not supported by observational studies using rumen pH measures possibly reflecting limited sample sizes, but also the weaker association of rumen pH with SRA.
Postulate 4) the number of new cases of disease should be significantly higher in those exposed to a supposed cause than in those not so exposed, as shown by prospective studies This postulate is based on the outcomes of prospective cohort studies.We did not identify any studies conducted using that trial design for the BAM nor studies using rumen pH AUC or cut points.Humer et al. (2015) studied the adaptation to lactation and a lactation diet in "SARA susceptible" and "SARA resistant" cows and found that AUC for pH of < 5.8 was greater for the susceptible cows but no difference in milk production or DMI was present.Similarly, Gao and Oba (2014) evaluated responses to a diet containing 35% and 65% concentrate in "susceptible" and "tolerant" late lactation Holstein cows, but despite an increased time spent under the curve for pH of < 5.8, total VFA concentration and profile did not differ between the groups, nor did DMI or milk production or milk fat, but susceptible cows sorted rations more.Neither of these studies provide support for use of a cut point for rumen pH < 5.8 to define SRA.Golder et al. (2014c) demonstrated that the effect of exposure to increased DMI of concentrates resulted in increased categorization of cows as being acidotic during a day with increased intake of concentrate (Figure 1).Cows with the greatest exposure, approximately 12 kg of ground wheat intake, were acidotic at 80% of sampling times and had significantly reduced milk and milk fat production.Stefańska et al. (2017) found that ruminal pH decreased when the dietary peNDF > 1.18 mm to starch ratio was reduced and, similarly, tended to linearly increase propionate, and valerate concentrations.Khorrami et al. (2021) evaluated responses in a meta-analysis of studies utilizing indwelling rumen pH probes to associate dietary factors including starch content, NDF, peNDF, and forage particle size to mean ruminal pH.Mean herd ruminal pH responses were not linear and the researchers identified responses for starch and a quadratic effect of peNDF with particle size > 8 mm once thresholds of > 20% starch and < 15% of peNDF were exceeded.Valentine et al. (2000) fed 7, 11, and 13 kg of concentrate (74% barley and 26% lupins) to cows fed on ryegrass and clover pasture and ryegrass silage.Milk (L/d), and protein yield (kg/d) and percentage increased for the greatest amount of supplement, whereas milk fat percentage but not yield, decreased.Bodyweight and BCS were greater for the cows receiving the greatest amount of supplement and BCS increased linearly with increased supplement.Ruminal propionate and valerate increased with supplement and acetate and butyrate decreased with supplement, while rumen pH did not significantly differ with supplement.Zhang et al. (2017) fed diets containing 20: 40: 60: and 80% concentrate to heifers and found linear increases in propionate, a cubic change in valerate and iso-valerate and a linear decline in rumen pH with increased concentrate amount.In an intriguing study, Tayyab et al. (2022) evaluated forage type, grass versus maize silage, and concentrate source, ground maize versus wheat, that resulted in 4 isonitrogenous diets differing in amounts of starch and NDF arising from different forage and concentrate sources.Effects in the study are not entirely clear as amounts of starch and NDF were not strictly controlled across comparisons but were analyzed as a factorial study.The effect of increased concentrate was to increase milk fat percentage attributed to a higher C16:0 content.Mean rumen pH and AUC < 5.8 did not differ with concentrate source and the only VFA concentration differences with concentrate were for isovalerate and acetate.McCaughern et al. (2020) increased starch from 14.9 to 22.5% and reduced NDF to 37.6 from 43.5% of the diet primarily through the addition of rolled wheat.The greater starch diets decreased mean pH and measures of AUC < pH 6.2, increased milk protein percentage, but did not influence milk fat percentage or yield, nor BCS over a 14-wk period.Gott et al. (2015) increased starch from 24 to 28% of the diet and challenged cows with a 32% starch diet, finding limited effects including a decrease in milk fat percentage with altered milk fatty acid profiles for the 28% starch group against the 24% starch and increased serum amyloid A for the 32% starch  2018) increased starch from 23 to 29% of the diet, finding increased milk protein percentage for higher starch intake, a similar milk fat yield, increased rumen propionate and lactate and decreased ammonia concentrations but no influence for mean pH, although pH AUC < 6.0 was increased by starch.Postulate 5 can be supported for the BAM.For rumen pH measures alone, the evidence is less compelling.

Postulate 5) a spectrum of host responses from mild to severe should follow exposure to a supposed cause along a logical biological gradient
Postulate 6) a measurable change in metabolism should appear regularly following exposure to a supposed cause in those lacking this response before exposure, or should increase in magnitude if present before exposure; this pattern should not occur in individuals not so exposed Table 1 provides a summary of studies demonstrating repeatable and measurable changes in metabolism.There is ample strong evidence to support this postulate.In general, there is also consistency in observations made using the BAM with rumen pH cut point studies; however, the evidence for decreased milk fat and milk fat to protein ratio may be stronger for the BAM than for pH cut points (Table 1).There is evidence for reduced Bacteroidetes and increased Firmicutes populations with increased exposure (Table 1).Increases in ruminal propionate and valerate concentrations are consistent for both evaluation methods and reduced ammonia concentrations with exposure has been consistently identified with the BAM.There is more evidence on increased LPS and inflammatory responses for the pH cut point and AUC studies than for studies using the BAM but good evidence overall for increases in concentration of histamine with exposure.An association of exposure with lameness was strong for Bramley et al. (2013) and supported by oligofructose challenge studies (Thoefner et al., 2004;Danscher et al., 2009), but not observational studies using pH cut points (Table 1).Associations to a lower BCS are mixed, some positive and some not, in the pH cut point studies and were not supported by Bramley et al. (2013).

Postulate 7) experimental reproduction of the disease should occur with greater frequency in animals or man appropriately exposed to a supposed cause than in those not so exposed
For experimental studies there is almost a circular argument for this postulate with studies using pH cut points or AUC and the BAM model providing ample evidence.The response to exposure with increased dietary intake of starches and sugars, differences in previous exposure to carbohydrate, alterations in fiber content or diet processing to decrease rumen pH and a nadir less than arbitrary threshold pH or AUC or increased concentrations of propionate, valerate, other VFA, and lactic acid and decreased ruminal ammonia concentrations are shown in Table 1.Golder et al. (2014c) demonstrated the effect of increased exposure to DMI of concentrates in increased categorization of cows as being acidotic during a day.Thoefner et al. (2004) fed oligofructose at 1.3, 1.7, and 2.1% of BW and markedly depressed rumen pH to < 5.0 for all treated heifers and found laminitis in 4 of 6 treated heifers.Similarly, Danscher et al. (2009) fed 1.7% of BW oligofructose to heifers and found rumen pH < 5.0, anorexia, depression, profuse diarrhea, and laminitis in treated heifers.Both these studies and that of Golder et al. (2012b), Golder et al. (2014b), andGolder et al. (2014a) who used experimental exposure to 0.8 to 1% of BW of starch and 0.2 to 0.4% of BW of fructose which created both increased SRA and CRA.This postulate is comprehensively supported.

Postulate 8) elimination (e.g., removal of a specific risk factor) or modification (e.g., alteration of a deficient diet) of the supposed cause should decrease the frequency of occurrence of the disease
While studies have been conducted that increased exposure and then reduced exposure, the potential for long-term damage to rumen papillae (Steele et al., 2011) and gastro-intestinal function, liver damage including abscess and other detrimental effects suggest that evidence from studies using modification of the diet are more compelling to evaluate this postulate.Further, cattle show cyclic eating behavior that reduces disease risk after exposure (DeVries et al., 2014;Golder et al., 2014b).Figure 1 demonstrates that an increased challenge results in a greater proportion of the day with SRA and that ruminal pH is also reduced.Changes in concentrations of valerate and propionate are consistent with increased risk of SRA with increased ingestion of rapidly fermentable carbohydrates.Golder et al. (2014c) also found a marked reduction in risk of SRA when canola meal was substituted in the diet for wheat.Table 5 provides details from studies with a focus on dairy cattle that modified rumen function by the addition of buffers or other rumen modifiers.Where suitable meta-analyses were available, these were used to evaluate effects (Table 5).The studies in Table 5 represent a subset from the literature that provided information on some of the more commonly used agents with substantial efficacy data on performance and ruminal metabolism.In toto, the findings of the studies provide support for Postulate 8.
Postulate 9) all relationships and associations should be biologically and epidemiologically credible All the relationships and associations are biologically and epidemiologically supported; however, we consider that the integrated BAM measures that include rumen pH, lactic acid, ammonia, and VFA concentrations more clearly represent the strength of association between exposure and the metabolic and disorder outcomes that are associated with SRA.This is evident in Table 4 that provides estimates of cut points, sensitivity, specificity, and AUC Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW for individual measures evaluated in the studies of Bramley et al. (2008) and Golder et al. (2023a).A marked difference is observed when the different VFA and rumen pH are tested for sensitivity and specificity and AUC; valerate has a sensitivity of 0.92 and specificity of 0.86, and an AUC of 0.89, propionate 0.95, 0.86, and 0.88, respectively; rumen pH (stomach tube 0.66, 0.74, and 0.70, respectively) and rumenocentesis pH (0.79, 0.71, and 0.75, respectively) with cut points of pH at 6.65 and 5.94, respectively (Table 4).

SUPPORTIVE DIAGNOSTIC MEASURES
Given that rumen pH measures are not highly diagnostic and VFA require laboratory support and are not readily available, other measures including clinical examination, other ruminal measures, blood, urine, fecal, milk, and other monitoring measures can be used to support diagnosis and monitoring of ruminal acidosis.Apart from the measurement of ruminal pH and milk fat to protein ratio, the diagnostic measures reviewed here are currently limited to use by research scientists due to their processing times but improvements in microsensors could become possible.
Diagnostic measures at the cow and herd level that can be easily, efficiently, and cost effectively conducted are needed.The use of tests in series can increase the specificity of diagnosis to reduce false positives and testing in parallel to increase the sensitivity of the test to reduce false negatives.Consequently, astute testing strategies can help identify the origin of the breakdown in management that precipitated ruminal acidosis.Standardization of methods for measuring indicators of ruminal acidosis is essential; in particular, collection site of rumen samples, sample storage, and collection time relative to feeding should be consistent.

Physical Examination, History and Clinical Signs
Cattle with CRA often have a dull demeanor, reduced activity, decreased rumination, be dehydrated, may seek roughage, may be lame, and reluctant to be moved to a chute for physical examination (Church, 1976).Active laminitis can be concurrent; however, the presence of laminitic rings on the hooves are signs of previous bouts of ruminal acidosis, and do not indicate current ruminal acidosis.Arthritis and polysynovitis have been associated with experimentally induced ruminal acidosis (Hyldgaard- Jensen and Simesen, 1966;Hidalgo et al., 2019).Rupture of minor pulmonary arteries into the bronchi and epistaxis, although rare, are clinical signs of sequalae of ruminal acidosis (Enemark, 2008), thus also do not indicate current ruminal acidosis.
Rumen fill may be low if cattle are inappetent.In CRA, scouring is common, and the feces may contain bubbles and have a sweet smelly acidic odor (RAGFAR, 2007) which may later change to foul-smelling that may be foamy and contain blood (Dirksen, 1970;Underwood, 1992).Table 1 indicates that there is not a strong association with SRA, possibly because diarrhea is multi-causal.In CRA poor body condition (Dirksen, 1970), a decrease in body condition since last examination, or poor body condition relative to herd mates of equivalent lactation stage may be observed.In CRA, a high pulse rate, ataxia, and low skin temperature may occur (Church, 1976) as may signs of abdominal pain such as kicking at the abdomen, laying in lateral recumbency or restlessness.
Cattle with CRA may have decreased rumen motility or rumen stasis (Dirksen, 1970).The rumen contents may feel watery and may splash on ballottement (Dirksen, 1970).The history, evidence of exposure to increased amounts or more fermentable carbohydrates and milk records are of great value during the clinical examination.These often provide definitive diagnostic support in CRA and strong support for a diagnosis of SRA.
By definition, SRA is not characterized by clinical signs; however, a high prevalence of lameness including laminitic rings, a history of epistaxis in the herd, poor body condition and low milk fat to milk protein ratio all indicate the potential for risk of SRA.

Dry Matter Intake and Feed Analysis
A reduction in DMI is a major response observed during ruminal acidosis (Britton and Stock., 1986).DeVries et al. ( 2014) evaluated a severe challenge with ruminally infused crushed barley fed at 10% of BW and demonstrated that increased time with a rumen pH < 5.5 resulted in increased feed sorting.Subsequently, cows resume eating in the recovery phase and again relapse, resulting in a cyclic feeding pattern (DeVries et al., 2014;Golder et al., 2014b).In parlor feeding systems, uneaten concentrate from a cow with ruminal acidosis is available for the subsequent cow that enters that milking bail and increases the risk of ruminal acidosis if that cow consumes more than her allocated concentrate.There is the potential to detect a decline in DMI in individual cows with monitoring sensors (Silberberg et al., 2024); however, in commercial TMR feeding systems detection otherwise is extremely challenging.Decreased TMR consumption at the herd level is unlikely as individual cows may have different risk states for ruminal acidosis at any time.Monitoring DMI in experimental facilities is considerably easier as feed intake is often measured.

Rumen Fluid Measures
The use of rumen fluid in clinical diagnosis began in the 1950s.Rumen fluid can be collected via stomach tube, rumenocentesis, or fistula.For CRA, a milky to milky-gray color can indicate ruminal acidosis and the fluid often contains grain.An acid smell is present during CRA (Rosenberger, 1979) and when rumen fluid is inactive a muddy to earthy odor is often present.Normal rumen fluid is slightly viscous and watery fluid suggests an inactive rumen (Rosenberger, 1979).Bubbles in rumen fluid suggest ruminal acid or foamy bloat.

pH
Ruminal pH can be measured by pH indicator strips or paper, pen testers (pH meter, display, and electrode in a single unit), hand-held meters, benchtop meters, or indwelling continuous boluses to various decimal points.Calibration of pH meters is critical.More critically, in terms of diagnostic potential, a highly accurate measurement of rumen pH is nearly impossible (Table 4).Figure 4 derived from Bramley et al. (2008) shows the correlations between rumen samples drawn by stomach tube and rumenocentesis in 660 cows (R 2 = 0.2).When the data from Bramley et al. (2008) were combined with those of Golder et al. (2023a) (Table 4), the stomach tube ruminal pH cut point was 6.65, the sensitivity 0.79, specificity 0.71, and AUC 0.75 indicating moderate diagnostic value.Rumenocentesis pH had similar diagnostic value to stomach tube when using the data from Bramley et al. (2008).Consequently, the diagnostic values for pH whether determined by rumenocentesis or stomach tube are similar (Table 4) and are not highly specific, nor sensitive.However, if used in conjunction with diet evaluation and clinical assessment, pH determinations provide support for a diagnosis (Supplementary Table 3; Golder and Lean, 2024).

Volatile Fatty Acids
The VFA have diagnostic value for SRA (Table 4; Figure 3) as indicators of rumen perturbation and ruminal acidosis and reflect responses to feed substrates and feed management.Measurement is usually by gas chromatography.In more severe cases of ruminal acidosis, CRA, total VFA concentrations decline (Wilson et al., 1975;Nagaraja and Titgemeyer, 2007) and complete absence of propionate, butyrate, and valerate and low concentrations of acetate can occur within 4 to 24 h of exposure to readily fermentable carbohydrates (Ryan, 1964;Kezar and Church, 1979;Golder et al., 2014a).Ruminal VFA concentrations returned to normal within 55 h of diet ingestion in a heifer that had recovered from CRA (Golder et al., 2014a), supporting the merit of ruminal VFA measurement as an indicator of CRA.When integrated with other rumen measures including ammonia, pH, and lactic acid, the VFA are very diagnostic, but both valerate and propionate have high diagnostic value as a single measure.

Ammonia
Measurement of ruminal ammonia concentrations can give an important indication of rumen function and bacterial activity.Measurement methods include enzymatic, colorimetry (spectroscopy), or titration.Ruminal ammonia is a useful diagnostic indicator of ruminal acidosis when used in conjunction with other indicators of ruminal acidosis (Table 4).

Lactic Acid
Ruminal lactic acid measurements may have value in the diagnosis of ruminal acidosis when cattle are fed diets with a high sugar content or are abruptly exposed to high-grain, rapidly fermentable diets.However, owing to the inter-conversion between stereoisomers, both lactic acid stereoisomers should be measured.Concentrations of lactic acid should be interpreted in combination with concentrations of VFA and ammonia, pH, and clinical signs.
The role of the D-and L-stereoisomers of lactic acid in ruminal acidosis remains unclear, with different proportions of the stereoisomers reported in different studies (Hibbard et al., 1995;Golder et al., 2014b).It has been suggested the ratio of stereoisomers may be influenced by ruminal pH (Giesecke and Stangassinger, 1980) and the D-stereoisomer was metabolized at approximately one third of the rate of the L-lactic acid stereoisomer in ruminal epithelial tissue slices (Prins et al., 1974).Thus, the D-stereoisomer is often considered to be more detrimental than the L, hence the term, D-lactic acidosis.The stereoisomers can also be interconverted by racemase; hence, their ratio does not always reflect their production (Asanuma and Hino, 2002).
Ruminal lactic acid reference ranges may need to be distinct for sugar and starch dominant diets given marked differences in response.Marked variation among animals was also observed in total lactic acid concentrations for grain and fructose challenged dairy heifers by Golder et al. (2012b) and by Golder et al. (2014b) suggesting individual animals respond differently and could have different tolerances for lactic acid, perhaps based on their rumen bacterial community or adaptations to diet.Transient spikes in ruminal lactic acid concentrations exceeding 15 mM have been reported through a day but were not influenced by concentrate content or buffer (Kennelly et   , 1999).The rapid flux of lactic acid under different conditions adds to the challenge of interpreting ruminal lactic acid results and target sampling timeframes; it could be easy to miss a spike in lactic acid with a spot sample.While many intervention and observational studies have not detected lactic acid we consider this to be a useful assay that can add insight.

Histamine
Given that the net absorption of histamine is low and it is inactivated either during or after absorption (Aschenbach and Gabel, 2000), it appears that histamine generated in the rumen may be less likely to cause laminitis associated with ruminal acidosis than endogenous sources of histamine (Brent, 1976).Therefore, the value of ruminal histamine concentration as a diagnostic measure for ruminal acidosis remains equivocal and reference values need to be established.

Lipopolysaccharides or Endotoxins
There are several different methods for testing LPS, but the Limulus Amebocyte Lysate (LAL) assay is generally considered the gold standard.Although ruminal and, in some cases blood LPS concentrations, may be associated with ruminal acidosis and nonspecific acute phase responses (Gozho et al., 2007;Khafipour et al., 2009a;Zebeli et al., 2012) use of LPS as a diagnostic measure for ruminal acidosis should be approached with caution until an involvement in the pathogenesis of ruminal acidosis is further elucidated, differences in LPS responses to different Gram-negative bacteria are known, and standard measurement protocols and reference values for cattle are validated (Monteiro and Faciola, 2020).

Bacteria
In the recent past, metagenomic (Li, 2015;Wallace et al., 2015;Zhao et al., 2020) and/or metatranscriptomic (Li and Guan, 2017;Ogunade et al., 2019) approaches toward investigation of rumen microbiology had been adopted but often lacked integration with metabolomics, other omic technologies, or clinical/ biological findings.At present, multi-omic approaches which involve the integration of multiple omic technologies (ie.metagenomics, metatranscriptomics, proteomics, and metabolomics etc) are being utilized allowing a more holistic interrogation of a biological system (Xue et al., 2020;Dai and Shen, 2022).
Researchers are also now supporting multi-omics findings with clinical/ biological measures such as epithelial metrics or molecular changes (McCann et al., 2016;Mu et al., 2022), improving validity of omics results.Plaizier  (Neubauer et al., 2020;Petri et al., 2021).While the majority of rumen acidosis microbiota publications have focused on communities in the rumen fluid, epimural bacteria have also been studied (Petri et al., 2013a;Wetzels et al., 2017;Li et al., 2019).Despite the technologies available, a large percentage of the ruminal microbiota and their functions are yet to be identified which is important to consider if attempting to use rumen microbiology to diagnose ruminal acidosis.It is possible that not all key causual microbes have been identified.There are novel opportunities to provide new prespectives on understanding not only the rumen microbiome but also the pathogenesis of ruminal acidosis and subsquently diagnosis.For example, Chen et al. ( 2021) performed rumen fluid transplants from healthy and SARA induced goats into antibiotic-pretreated mice and induced colonic inflammation and similar fermentation patterns and colonic microbiota to the donors.
In general, Gram-negative bacteria predominate in forage-or roughage-fed cattle with increases in the number of Gram-positive bacteria observed as concentrate feeding increases (Hungate, 1966;Latham et al., 1971) and the rumen environment may adapt to feed changes over time.Gram-negative bacteria appear to be replaced with Gram-positive cocci and rods during ruminal acidosis (Dirksen, 1970;Nagaraja and Titgemeyer, 2007;Zhang et al., 2020).Bacterial taxa function as communities, rather than single taxa, making it a challenge to assess whole community shifts.A large amount of redundancy is present, likely providing resilience to the host during rumen challenges.Taxis et al. (2015) reported ruminal ecosystems that were dissimilar in their taxonomy to have similar metabolic functions.This serves as a caution that there is not likely to be a singular rumen microbial taxonomic profile that should be considered diagnostic for ruminal acidosis.
Individuals appear to have a unique rumen ecosystem comprised of a core and non-core rumen microbiome (Jami and Mizrahi, 2012;Henderson et al., 2015;Xue et al., 2018) that has a unique ability to adapt to different feed substrates (Dougherty et al., 1975;Brown et al., 2000) and may be associated with an individual animals' susceptibilities to disorders such as ruminal acidosis.The core taxa, i.e., microbes that are shared between individual animals (Wirth et al., 2018) are likely critical for basic function while the non-core increase resilience (Li, 2015).Many have attempted to define the core taxa (Li et al., 2012a;Wu et al., 2012;Petri et al., 2013b;Wirth et al., 2018;Xue et al., 2018) at various taxonomic levels and are likely to continue to do so as reference databases grow.A rumen transplant study showed the core microbiome was not affected during transplantation but more work is required to determine transplantation sensitive species (Mu et al., 2021).Understanding the dynamic nature of both the core and non-core microbiota will assist both diagnosis and control of ruminal acidosis.Golder et al. (2023b) identified that the odds of center logged tranformed relative abundance of the Firmicutes and Spirochaetes phyla increased in cattle from Australia, USA, and Canada with a high-risk of acidosis, the odds of Lentisphaerae, Planctomycetes, and Tenericutes decreased.The area under the curve for this model was 0.925 showing it was strongly associated with high-risk of acidosis.Consequently, diagnostic models based on bacterial analyses of rumen fluid can be expected, but as noted earlier are not likely to be singular models.Perhaps real time monitoring of rumen microbiota, possibly with the use of artifical intelligence will be possible in the future.Artifical intelligence has also already been used to show that rumen microbiome composition explains a significant portion of the variation in residual feed intake (Monteiro et al., 2024), despite residual feed intake being a flawed measure (Old et al., 2024).

Host genetics
Genetic associations with ruminal acidosis category have been investigated in 2 small studies (Golder et al., 2018;Golder et al., 2023c); however, Golder et al. (2023c) only found a tendency for an association with the probability of being in the low-risk ruminal acidosis category.It is very possible that the limited sample sizes precluded significant associations.
Host genetics may play a vital role in rumen bacterial profiles as bacterial profiles are related to breed (Guan et al., 2008;Brulc et al., 2009) and breed of cattle had a greater influence than diet on bacterial profiles (Lee et al., 2012).Further, Weimer et al. (2010) showed that when > 95% of ruminal fluid with differing pH, total VFA concentration, and bacterial community composition from 2 cows fed the same diet was exchanged the ruminal pH, total VFA, and bacterial profiles of the 2 cows returned to their original profiles within 24 h.Subsequent rumen transplant work supports the theory of a host specific microbiome; Mu et al. (2022) showed the core microbiome was not influenced by transplantation as it is already highly compatible, while Zhou et al. (2018) showed microbial re-establishment was highly individualized after transfaunation indicating that there is considerable uniqueness in the interaction between the mammal and associated microbiota.Associations between the host and the relative abundance of individual bacterial phyla and families were found in a pilot study (Golder et al., 2018) and a 293-cow study (Golder et al., 2023c) but greater numbers of observations are needed to characterize these Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW relationships.Host specificity, of ruminal bacteria, even in the non-core may pose a challenge when examining interventions or management changes that effect the rumen and emphasizes the need for large sample sizes for in vivo studies.In particular, the variation in ruminal bacterial communities among cattle will pose challenges for strategies designed to control and diagnose ruminal acidosis.

Milk Measures
Table 4 suggests milk fat to protein ratio has some merit as an indicator for ruminal acidosis with a sensitivity and specificity of 0.54 and 0.82, respectively and AUC 0.68 (Table 4).Milk fat percentage and yield can be affected by other factors apart from ruminal acidosis including stage of lactation, breed, ration composition, and body fat mobilization (Grummer, 1991).Further, milk fat content had low correlation coefficients with ruminal pH in cows > 30 DIM with r = 0.305 (Allen, 1997) and r = 0.390 (Enemark et al., 2004).Cows < 30 DIM had an r = -0.06(Enemark et al., 2004) and a regression from the Bramley et al. (2008) and Golder et al. (2023a) data combined had an r 2 of 0.01 showing that milk fat measures alone are not indicators of ruminal pH for cows in the first 100 d of lactation.Kleen et al. (2003) concluded that milk fat depression appears to occur in the same situations as SARA but might not depend on the presence of SARA.Specific milk fatty acids may be better indicators of risk (Mensching et al., 2021).Colman et al. (2010) found that control and acidotic cows could be discriminated using iso-C13:0, iso-C16:0, and C18:2 cis-9,trans-11, while Gott et al. (2015) found that cows fed a high-starch diet had greater concentrations of C18:1 trans-10, C18:2 trans-10,cis-12, and most oddchained fatty acids and lower concentrations of most branch-chained fatty acids than controls.Differences in the fatty acids identified as diagnostic in these and other studies will reflect difference in diets and possibly herd genetics and indicate likely limitations in the application of fatty acid profiles as monitoring tool.While milk fat to protein ratio is specific (0.81), it should only be used in combination with other indicators of ruminal acidosis to diagnose ruminal acidosis.Other milk markers that have also been associated with ruminal acidosis include lactose, chloride, sodium, potassium, and milk urea nitrogen (Enemark and Jorgensen, 2002), but all require further validation.

Blood Measures
Blood measures are not routinely used for diagnosis of ruminal acidosis.Dirksen (1970) noted, later supported by Brown et al. (2000), that blood composition constantly changes as ruminal acidosis progresses; many of the changes reported are from peracute cases.Thus, blood measures are not a reliable diagnostic tool.Nagaraja and Titgemeyer (2007) provide a summary table of the blood changes in clinical versus subclinical cases.During CRA blood analysis will generally show an increase in packed cell volume with hemoconcentration as a result of an increase in the increased osmolarity of the rumen contents due to an accumulation of VFA and glucose (Hernández et al., 2014).Total blood protein, urea, non-protein nitrogen, and total bilirubin may also be increased (Dirksen, 1970;Church, 1976).Leukocytosis with neutrophilia can reflect stress and anemia, potentially, from ruminal ulcers and hyporexia (Hernández et al., 2014).Low blood pH, base excess, and bicarbonate are expected in CRA (Nagaraja and Titgemeyer, 2007;Hernández et al., 2014).Blood chloride concentrations may drop before death, while glucose will rise (Church, 1976).Blood lactate and inorganic P will increase in CRA (Church, 1976).Lipopolysaccharides, inflammatory mediators, and acute phase proteins may also increase (Gozho et al., 2007;Khafipour et al., 2009b; a; Table 1); however, these are non-specific indicators of inflammation that can be highly variable and should be interpreted in combination with other indicators of ruminal acidosis (Aschenbach et al., 2019).Brown et al. (2000) produced a series of models to classify steers experiencing acute acidosis, SARA, or not affected by a carbohydrate challenge based on blood and rumen measures, DMI, and rectal temperature over several days of sampling.These provided a series of inconsistent measures between comparison groups and days.Marchesini et al. (2013) evaluated use of 17 blood, blood gas, hematological, and acute phase proteins to classify heifers experimentally fed control, low, medium, and high starch diets into 4 ruminal pH groups: normal, risk of acidosis, SARA, and acute ruminal acidosis using canonical discriminant analysis.They found using the combination of hemoglobin, mean platelet volume, β-hydroxybutyrate, glucose, and reduced hemoglobin was discriminatory (Marchesini et al., 2013).While single blood metrics appear unreliable, panels of tests may be beneficial for diagnosis of ruminal acidosis but require more investigation.

Sensors
The integration of sensing monitors using visual, indwelling ruminal pH, temperature, and motion detectors (ruminal and mastication), and activity detectors provide potential for detection of disease and routine monitoring of metabolism remotely (Bewley et al., 2008;Beauchemin, 2018;Hamilton et al., 2019;Dijkstra et al., 2020;Knight, 2020;Han et al., 2022).Further, there has been integration of these findings with routine milk testing using mid infrared reflectance spectroscopy (MIRS) (Mensching et al., 2021) to detect "ruminal acidosis."Activity monitors are being increasingly used to provide health alerts to dairy personnel.These alerts result from algorithms that detect when a cow's activity deviates below her mean activity level, thus they are non-specific health alerts (Steensels et al., 2017;Antanaitis et al., 2019;Rial et al., 2023).Sensor signals are influenced by external factors, for example rumen temperature is influenced by drinking events.Further, ruminal pH detection systems have been noted to drift over time (Dijkstra et al., 2020) and require careful statistical analysis due to marked variations in ruminal environments among cattle (Denwood et al., 2018;Dijkstra et al., 2020).
Rumination time has been used as a method of monitoring ruminal acidosis with a 37% shorter chewing time observed for affected cattle (Antanaitis et al., 2024).The recommended percentage of cows chewing their cud is generally considered to be around 40% (Eastridge, 2000;Maekawa et al., 2002) to 50% (RAGFAR, 2007).Cattle with "SARA" had less time spent eating in a study where starch in the diet was markedly increased from 10.5 to 31.5% (Silberberg et al., 2024).Substrate, therefore, influences rumination and while a decline in rumination time is non-specific for ruminal acidosis (Steensels et al., 2017;Rial et al., 2023), there appears to be potential to apply these methods in combination with observations on milk yield, milk fat to protein ratio or MIRS, especially based on milk fatty acid profiles (Mensching et al., 2021;Silberberg et al., 2024) to increase its diagnostic value.Han et al. (2022) review the requirements for effective real time monitoring of biomarkers in the rumen along with the current research on various types of biosensors and their challenges.There is limited experimental work on sensors that can detect VFA from the breath.Sensors that could be placed in the entry or exit of the parlor, drinking points, lying stalls, or feeding lanes or in the boluses in the rumen have potential for high diagnostic value.

Other Measures
Urine (Furll, 1994), fecal (Oetzel, 2000), ruminal gases (Dewhurst et al., 2001), blood gases and electrolytes (Enemark and Jorgensen, 2002), inflammatory measures in synovial fluid (Hidalgo et al., 2019), and metabolomic profiles (Saleem et al., 2012) can be used as indirect indications of ruminal acidosis.Urine pH may fall; specific gravity may increase with excess excretion of inorganic P, protein, bilirubin, ketone bodies, lactic acid, and glucose during acute ruminal acidosis (Hyldgaard-Jensen and Simesen, 1966).Oxidative stress responses have been influenced by concentrate diets (Gabai et al., 2004;Wullepit et al., 2009); however, responses were not observed when cattle were fed a single challenge feed of carbohydrates (Golder et al., 2013) or in a long-term carbohydrate challenge study (Golder et al., 2014b).Further research is required to investigate the potential of oxidative stress responses as indicators of ruminal acidosis.Rumen fungi have important roles in fiber digestion (Hess et al., 2020) and protozoa in carbohydrate

CONCLUSIONS
Ruminal acidosis occurs as a continuum of disorders of varying severity.Clinical ruminal acidosis is an important condition of cattle and SRA is very prevalent in many dairy populations.Historical confusion about the etiology and pathogenesis of ruminal acidosis led to definitions that are not fit for purpose as acidic ruminal conditions solely characterized by ruminal pH fail to reflect the complexity of the condition.Use of a model, such as the BAM, based on integrated ruminal measures, for evaluating ruminal acidosis is fit for purpose but requires a method to simply apply in the field.While it is likely that this model will be refined and other models that provide more frequent measures may suit certain hypotheses better, the critical value in the model is that it demonstrates that ruminal acidosis is much more than rumen pH and that performance of cattle is much more closely aligned to a model that considers more than rumen pH.Even with the use of such a model, astute evaluations of the condition whether in experimental or field circumstances will be aided by ancillary measures that can be used in parallel or in series to enhance diagnosis and interpretation.Promising pathways for future diagnostics include multi-omic evaluations of inflammatory pathways and ruminal microbial populations and real time sensing of cattle behavior and performance using motion, intra-ruminal and chemical sensors.

NOTES
This study received no external funding.The authors wish to thank N. Jonsson (Institute of Biodiversity, Animal Health and Comparative Medicine, University of Glasgow, Glasgow, United Kingdom) and C. Knight (CKP BreatheScience, Scotland) for insights.Also F. Mulligan and researchers at University College Dublin (Ireland) for making their data available.The authors also wish to thank Jan Troutt for assistance with Figure 2, D. Sheedy (Scibus) for assistance with producing Figure 3, and K. Hobson and J. Tawyer (The University of Sydney, Camden, NSW, Australia) for assistance with literature searching of dictionary definitions of disease states and referencing, respectively.

Figure 3 .
Figure 3. Receiver operator curves for the diagnostic value of rumen valerate, propionate, caproate, D-lactic acid, acetate, butyrate, and ammonia concentrations and pH measured in rumen fluid collected with a stomach tube.The dashed line is the reference line.Data were sourced from samples obtained by Bramley et al. (2008) and Golder et al. (2023a)(n = 1,086).
Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW challenged cows.Dias et al. ( Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW
Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW metabolism(Rosenberger, 1979), hence they are likely to be useful indicators of ruminal acidosis as technology develops.

Table 3 .
Golder (2014)d relationship between ruminal pH measurements in ruminal fluid collected using stomach tubing, rumenocentesis, and rumen fistula methods in cattle.Sourced fromGolder (2014) 1 Difference in ruminal pH values were calculated by subtracting the mean ruminal pH value for the second named ruminal collection method from the first named collection method i.e., Mean ruminal pH of stomach tube ruminal sample -Mean ruminal pH of rumenocentesis ruminal sample. 2 Mean over 1 min.3Meanover 5 min.

Table 4 .
Bramley et al. (2008) and area under the curve, and cut-off points from receiver operator curves for the ruminal acidosis diagnostic value of rumen and milk measures from samples obtained byBramley et al. (2008) and Bramley et al. (2008) two measures based on sensitivity and specificity: their product (Liu index); their sum (Youden index), and the point on the curve closest to sensitivity = 1 and specificity = 1 were estimated with cutpt (Stata v18.0,StataCorp,Tx). 2 Rumenocentesis samples are only from samples obtained byBramley et al. (2008)(n = 707).3 n = 861.

Table 5 .
Summary of interventions to control ruminal acidosis in dairy cattle with a focus on production, ruminal, and metabolic changes consistent with prevention of ruminal acidosis Intervention

Table 5 (
Continued).Summary of interventions to control ruminal acidosis in dairy cattle with a focus on production, ruminal, and metabolic changes consistent with prevention of ruminal acidosis Continuedal.

Table 5 (
Golder and Lean: RUMINAL ACIDOSIS: A CRITICAL REVIEW Continued).Summary of interventions to control ruminal acidosis in dairy cattle with a focus on production, ruminal, and metabolic changes consistent with prevention of ruminal acidosis et al. (2017) investigated hindgut communities in parallel to ruminal communities in a ruminal acidosis study while others are focusing on hindgut communities NS; Not significant at P > 0.05.