The concept that the modern diet produces excess acid, which causes several diseases of modern societies, and that "alkaline diets" prevent and cure these diseases are marketed to the general public across the globe. The public is being encouraged to measure their urine and/or salivary pH to assess their health status and their risk of disease 1 2 3 4 . Marketers claim that alkaline diets and related commercial products counteract acidity, help the body regulate its pH, and thus prevent disease processes including osteoporosis, cancer, and cardiovascular disease through websites, (e.g. 1 2 3 4 ) flyers, magazines, direct mail marketing, and books 5 6 7 8 directed to lay audiences. A Google search of "acid ash diet" and "alkaline diet" resulted in 1.4 million and 400,000 hits respectively. As well, the acid-ash hypothesis has been broadly stated as a major modifiable risk factor for bone loss in osteoporosis in well cited scientific papers 9 10 , textbooks 11 , reference work 12 , a government-funded workshop summary 13 , and lay literature.

According to the acid-ash hypothesis, high dietary protein intakes are detrimental to bone health since protein is an important "acid generating" diet component, and structural bone mineral is dissolved to release bicarbonate to neutralize acid and avoid systemic acidosis 9 14 15 16 . A recent narrative review claimed: "acid-yielding diets (cereal grains and most dairy products) cause urinary calcium loss [and] accelerated skeletal calcium depletion..." 17 . Dietary protein associated increased urinary calcium has been considered confirmation of this theoretical effect 15 18 19 20 21 .

Some critical reviews of the acid ash hypothesis have been undertaken with regards to bone health (in terms of the biochemistry 22 23 24 25 , the role of protein 26 27 , and phosphate 28 , calcium balance 29 , and the hypothesis in general 30 ), however, to our knowledge, no systematic review has been done to assess the strength of the evidence of the acid ash hypothesis in terms of the etiology of osteoporosis.

The purpose of this systematic review was to evaluate causal relationships between the dietary acid load and osteoporosis among adults, and to assess the evidence using Hill's Criteria. The specific objectives were to examine the evidence that lowering the diet acid load alters the risk of osteoporosis progression by: a) conducting a thorough search of the literature for randomized human intervention and prospective observational studies and in vitro bone studies, of the dietary acid load and osteoporosis; b) performing meta-analyses of urine calcium, calcium balance, changes of bone mineral density (BMD), fractures, changes in bone strength, and bone resorption marker (BR marker); c) evaluating the prospective observational studies for osteoporosis risk factors that were controlled and not controlled in the analysis; d) reviewing the in vitro experimental findings to determine the pH at which increased bone resorption took place. Additional objectives of this study were to examine the exposures, the purported detrimental aspects of the diet acid load, for internal consistency, that is whether the food and urine estimates of the hypothesis were consistent with the whole hypothesis. We used Hill's criteria of causation 31 to assist in the assessment of causal relationships between exposures and disease 32 33 , in this case, between the acid load of the modern diet and osteoporosis. Low quality trials are much more likely to demonstrate a benefit from an intervention 34 35 36 , thus experts recommend that systematic reviews report meta-analyses restricted to trials at low risk of bias either as the primary analysis or in conjunction with less restrictive analyses 36 . Therefore we evaluated the included studies for their risks of bias, and focused the meta-analyses on high quality studies.

The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Statement 37 was used to guide this study.

In an attempt to find all published literature on the topic, studies relating to the acid-ash diet hypothesis and bone health were identified through computerized searches using, but not limited to, the medical subject headings and textwords, first: acid, alkaline, acid-ash, acid-base, modern, western, diet, calcium, phosphate, acid-base equilibrium, acid excretion, net acid excretion, bone or bones, osteoporosis, urine, balance/retention, biopsy, fracture(s), bone mineral, and bone mineral density. Second, to find studies that have examined proposed mechanisms for the hypothesis, we used the terms hydrogen-Ion concentration, cells (cultured), and mechanism. Databases searched included Medline back to 1966 (PubMed), Cochrane Database of Systematic Reviews, CINAHL back to 1982, EMBASE back to 1980, and the Cochrane Controlled Trials Register, up to August 2010. A Librarian (DL) was consulted regarding the literature search. In an effort to include all available studies, reference lists were reviewed for additional relevant articles. The literature search was not limited to English language articles.

Article titles were examined for potential fit to the inclusion criteria by one reviewer (TRF). When the title was not clear regarding the potential fit, the abstract was reviewed; when the abstract was not clear regarding whether the study fit the inclusion criteria, the paper was reviewed. Authors were contacted for additional information. Two authors (SCT & TRF) independently rated the randomized studies for their risk of bias using the Cochrane Risk of Bias Tool 43 ; two (AWL & TRF) extracted the BR marker data; and two (ME & DAH) extracted the potential confounders controlled in the cohort studies. Differences of opinion were resolved by discussion to achieve consensus.

After the data, including exposures and outcomes, was extracted and described in tables, the risk of bias of the randomized studies was assessed using the Cochrane Risk of Bias Tool 43 , for the randomized studies of the acid ash hypothesis by SCT and TRF.

To address the questions of what evidence supports the acid-ash hypothesis for the role of net acid excretion (NAE) and phosphate in urinary calcium excretion and calcium balance, we used the highest quality of evidence available, meta-analyses of random control trials (RCTs) or random cross-over studies (RCO). When meta-analyses were not found that fit the inclusion criteria and randomized trials were found that did meet the criteria, then meta-analyses were performed. When the exposure was a continuous measure, then linear regression analyses, weighted for sample size, were used to combine the results from the included studies to examine the effect of NAE and dietary phosphate on urinary calcium excretion and calcium balance.

To examine the effect of an alkaline treatment or a reduced "acid" diet load on the changes of resorptive bone resorption markers (BR markers) (i.e. serum C-telopeptide (CTX), urine N-telopeptide, and urine deoxypyridinoline crosslinks) using meta-analysis techniques, the exposures were considered as alkaline versus control, and standardized mean differences were calculated using fixed and/or random effects models, with Cochrane RevMan5 (Version 5.0. Copenhagen, The Cochrane Collaboration, 2008). If the p-value for heterogeneity was between 0.05 and 0.5, then a random effects model was used 44 . The BR marker changes from baseline were used when baseline values were available in the RCTs, or the differences between the groups if baseline values were not available. Urine CTX was not included since it is considered less valid than serum CTX due to higher biological variation 45 . Then a second meta-analysis was performed as a sensitivity analysis, on the BR markers measured in a fasting state and at the same time of day as recommended to decrease measurement errors 46 47 . A difference greater than 30% for serum markers or 50-60% for urine markers were considered clinically important 46 47 .

The in-vitro cell culture studies of bone demineralization at varying pH were examined to determine the pH of testing and to summarize the effects revealed within the physiological pH range.

Prospective observational studies of the acid-ash hypothesis were examined for whether they supported or did not support the hypothesis, as well as which osteoporosis risk factors 48 49 50 51 52 were controlled for in the analysis.

Hill's criteria of causation 31 (Table 1) was used to evaluate the possibility of causation by the acid load of the modern diet on the etiology or potentiation of osteoporosis. Hill's criteria include: whether the exposure precedes the disease in time (Temporality), whether a dose-response or Biological Gradient relationship exists, the Strength of the evidence, whether the concept is Biologically Plausible, whether the evidence has Consistent findings between the studies of various designs, and whether actual Experiments have been done to determine whether altered exposure results in changes in disease frequency 31 .

This systematic review, based on quality randomized trials and prospective observational studies, did not find support for the acid-ash hypothesis which states that "acid" from the modern diet causes osteoporosis or that an alkaline diet or "alkaline" supplements or salts prevent osteoporosis. Applying Hill's criteria to this body of evidence provides additional insight into the likelihood of causality based on established criteria. The criterion related to the strength of association partially was met because the quantity of urine calcium was related to the diet acid load, however, calcium balance which is the preferred measure of calcium metabolism, was not related to diet acid load. The criteria for the biological gradient, biologic plausibility, consistency and experimental evidence for a casual relationship for the acid ash hypothesis were lacking with regard to whole body calcium balance, bone resorption markers, and changes in BMD. Thus, claims that the modern "acid" producing diet causes osteoporosis were not substantiated by research evidence. These finding suggest there is not likely any bone health benefit from consumption of commercial products intended to counteract this dietary acid.

Hill's criterion requires consistent evidence from a variety of experimental designs and the studies of food estimates of acid load remain inconsistent and show only associations with urine calcium rather than a causal relationship for osteoporosis. Although there is general agreement in the commercial literature and advertising about which foods contribute acid and base, the foundation of these statements is weak and it is unclear whether the calculated amounts of acid or base have any association with health or disease.

In addressing these findings within the context of the existing acid-ash hypothesis, limitations to the evidence can be identified. Evidence or limitations to the evidence arise in the following areas: 1) lack of support for the hypothesis by well-designed calcium balance studies; 2) lack of well-designed studies with more direct measures of this disease (bone strength as measured by fragility fractures or biomechanical testing of bone biopsy material); 3) lack of control of important potential risk factors among the longitudinal cohort studies; and 4) lack of a defined mechanism that could occur at physiological pH. Small alterations in the surrogate measures of calcium in the urine and/or changes in BR markers are not evidence that alterations in the diet acid load cause bone demineralization. Additionally biological plausibility is questionable because of the conflicting roles of phosphate, sodium, potassium, protein and calcium interactions, and milk, since the roles put forth by the hypothesis differ from the actual roles of these molecules with respect to osteoporosis.

In the acid-ash model, sodium is one of the cations that has been assumed to represent base excretion, and cations theoretically protect against bone calcium losses 15 . In the model of the acid-ash hypothesis, sodium is considered to have a similar bone protective effect to calcium, potassium, and magnesium. However, experts consider high sodium intakes to be a possible risk factor for bone mineral loss 48 . It is possible that the acid-ash hypothesis is over simplistic in its categorization of sodium, potassium, calcium and magnesium as protective ions vis-à-vis bone health.

Early work to define food sources of acid and base began early in the previous century 110 112 265 . Sherman published tables listing the acid and base contributions of 64 foods based on the ashed foods' mineral content in 1912 110 . Remer and Manz updated the original calculation in 1995 when they published newer tables of food acid loads with simple correction factors for each mineral designed to take imperfect absorption into account 152 . The 1912 and 1995 food lists share a premise that urinary excretion of hydroxides of sodium, potassium, calcium and magnesium reflect "base" excretion while urinary protonated forms of phosphate, sulfate, and chloride reflect "acid" excretion 110 152 265 . However, the assumption that food lists can reliably and exclusively classify foods as dietary sources of excreted acid or base is not supported by this review. The food lists categorize dietary phosphate and protein containing foods as acid sources anticipated to enhance bone loss, while evidence suggests that dietary phosphate does not increase calcium excretion (Figure 4) or decrease calcium balance (Figure 5), and dietary protein may enhance or protect BMD 263 264 . Although Remer and Manz estimated that milk has a slight acidic load 152 , other investigators estimated that milk would supply an alkaline load 110 122 , and a recent study revealed that milk actually contributes an alkaline load 207 . Grains were included in the food lists as acid generating, and have been considered "acid-yielding" 17 , but have not been evaluated for their hypothesized acidogenic and calciuric responses or effect on bone health, although two attempts have been made 133 197 .

The measurement of the diet acid load based on urine titratable acidity, ammonia and bicarbonate 14 15 115 is not a precise estimate due to problems with each constituent 24 . Ammonium (as ammonia) and bicarbonate (as CO₂) may be lost due to volatility 266 prior to their measurement. Additionally, the measurement of titratable acidity is influenced by poor solubility of calcium, phosphorus, and uric acid, which can cause an over or underestimation of titratable acidity 24 267 . Therefore the measurement of urinary acid excretion is error prone and may not accurately reflect the exposure to dietary acidity.

The majority of experimental evidence supporting the acid-ash hypothesis is derived from studies that have used urine calcium and/or BR markers as the outcome measure. Urine calcium changes are confounded by changes in calcium absorption. The estimated change in BR markers (Figures 6 &7) is less than the "least significant change" needed to consider that a true biological effect has occurred as opposed to a change due to measurement error 46 47 268 and the results are inconsistent. The better-designed RCT which used changes of BMD as the outcome did not support the hypothesis 68 . The lack of consistent information regarding the effects of protein, milk and grain foods on bone undermines support for the acid-ash hypothesis, and the unreliable measurement of acid excretion in urine further undermines the hypothesis (Table 5). Therefore, the experimental evidence does not support the acid-ash hypothesis (Table 5).

The acid-ash hypothesis recommends that to maintain bone health people consume generous quantities of fruit and vegetables (8 to 10 servings per day 145 152 242 ) along with modest amounts of grain and protein foods 21 42 . Generous quantities of fruit and vegetables are not likely to be harmful and may have other health benefits 269 . It is possible that fruit and vegetables are beneficial to bone health through mechanisms other than via the acid-ash hypothesis since there is some preliminary human and animal evidence that some fruits and some vegetables have supportive effects on bone 270 271 .

In contrast to the acid-ash hypothesis, recent research suggests that sufficient protein intake is needed for the maintenance of bone integrity 27 29 62 263 263 264 . Changes in urine calcium and BR markers should not be considered proof of the acid-ash hypothesis.

Based on the review of the literature to date and an application of Hill's criteria to the evidence the relationship between dietary acid with risk of osteoporosis is not confirmed. Further research is needed to determine whether fruit and/or vegetables are protective of bone health and what are the ideal protein intakes for bone health.

BMD: Bone mineral density; BR markers: bone resorption markers; CI: confidence interval; CTX: C-telopeptide; DPD: deoxypyridinoline; NAE: net acid excretion; RCT: randomized controlled trial;

The authors declare that they have no competing interests.

The author's responsibilities were as follows: TRF, SCT, & AWL designed the study, TRF, SCT, AWL & DAH extracted the data; TRF searched the literature, performed the statistical analysis and wrote the manuscript. ME directed the study's statistical analysis and graphic representations, SCT and TRF independently rated the randomized studies for their risk of bias using the Cochrane Risk of Bias Tool. All of the authors contributed to interpret the findings and writing the manuscript, and all authors read and approved the final manuscript.

This work was supported in part by: Doctoral fellowships from the University of Calgary and the Alberta Heritage Fund for Medical Research. The funders played no role in study design, collection, analysis, interpretation of data, writing of the report, or the conclusions reached.