This dataset contains information from the meta-analysis presented in "Acids in Coffee: A Review of Sensory Measurements and Meta-Analysis of Chemical Composition." Acid concentrations were extracted from a total of 121 publications for at least one of 26 different organic acids (OAs) or 23 different chlorogenic acids (CGAs), yielding 7,509 distinct data points. Concentrations were collected for Coffea arabica, Coffea canpehora (robusta), and other types of coffee, for both green and different roast levels. 

To obtain a more complete picture on the acid composition in coffee, we conducted an extensive review and meta-analysis of the scientific literature. Web of Science, Google Scholar, and the University of California Library catalog were searched between April to December 2020 for any publications that included data about the amounts of acid in coffee samples. This search focused explicitly on measurements of the concentration of individual CGAs and OAs in coffee, not the overall amount of acid in coffee (usually expressed as total titratable acidity). Access was limited to online versions of publications due to COVID-19 restrictions during the time of the database search. Articles not available directly online were obtained through Interlibrary Loan requests. In the case of articles published in languages other than English, a translating website was used to read the article. Abstracts and full texts were examined for specific data about the absolute amounts of any chlorogenic or organic acids. Articles that only examined the presence, relative amounts, or formation pathways of CGAs or OAs were excluded. Papers that reported CGAs or OAs in units of mg/L without including the original mass of coffee used were excluded due to the fact that amounts in units of mg/L cannot be directly compared with amounts in units of mg/kg (comparing mass in wet basis versus mass in dry basis). 

If the publication did contain specific amounts of CGAs or OAs that satisfied the preceding conditions, then all roast levels, extraction types, and coffee species were included, except for decaffeinated and instant coffee. The additional processing on decaffeinated and instant coffee complicates comparison with other coffees. If the publication listed data for store-bought samples, those were included as well. In some cases, roast level and coffee species were not specified, and these data points were categorized as “unspecified”. For the purposes of this review, Coffea arabica will be referred to as “arabica” coffee and Coffea canephora cv. robusta will be referred to as “robusta” coffee.  
A tremendous complicating factor is the roast level, which strongly affects acid concentrations but is very challenging to quantify precisely; subjective roast descriptions like “dark roast” have no universally accepted definition. For the purpose of the meta-analysis, we therefore performed a semi-qualitative classification of the reported roast levels into three categories – light, medium, or dark – using the following methodology.    

The roast levels for specific data in publications was determined in one of four ways: (1) as the publication’s self-described roast level; (2) from the publication’s reported amount of water lost during roasting (11-13% = light, 14-16% = medium, 17-20% dark) or organic roast loss percentage (ORL%) (2-4% = light, 4.1-5.5% = medium, 5.6-7% = dark) (Perrone et al. 2008; Weers et al. 1995); (3) the publication’s reported L*a*b* color values of the roasted beans where L* of 30, 25, and 20 correspond to light, medium, and dark, respectively (Chindapan, Soydok, and Devahastin 2019); or (4) as “unspecified” if the publication did not mention any of the above.  If the publication provided finer demarcations of roast level (e.g., a “light roast” and a “very light” roast), then we grouped their samples as appropriate into just our three broad categories. Lastly, samples that were labelled simply as “roasted” without giving any indication to the degree of roast kept the label of “roasted” and were included when comparing roasted coffee as a whole (Correia, Leitao and Clifford 1995; Agnoletti et al. 2019).   We emphasize that because roast level is very qualitative and methods of measuring roast level vary greatly, the roast level labels used in this paper are approximate, based on the information available in the cited publications. 

Similarly, extraction of the acids for analysis varied widely among the different publications. If a chemical solvent such as methanol was used, the extraction type was labelled as “solvent”; soaking the coffee grounds in hot water was labelled as “immersion”; extraction types such as “French press” or “espresso” were explicitly mentioned in their respective publications and the labels were kept for data collection. 

Lastly, all measurements were converted to mg/kg to simplify comparison.  Accordingly, the units reported in the publications often had to be converted, e.g., data reported in units of g/kg was multiplied by 1000 to match units of mg/kg. In cases, where publications reported concentration in terms of mmol/kg, the molecular weight of the specific acid was used to convert to mg/kg. Lastly, in articles that presented the data in units of mg/L and included the original brew recipe (grams of coffee and liters of water), the data was converted to units of mg/kg using the brew recipe, assuming full extraction from the dry coffee grounds. 
Data for 23 different CGAs and 26 different OAs was collected and analyzed While thirty-eight OAs have been quantified in coffee (Maier 1999), many are present in trace amounts and not commonly reported. Those reported in fewer than 2 publications and with amounts less than 0.01/kg were not included, accounting for the difference in total OAs analyzed in this review. 

In chlorogenic acids the widely reported acids are total CQA, 5-CQA, 4-CQA, 3-CQA, total diCQA, and total FQA. Some publications would report only total concentrations of one class (“Total diCQA”) instead of quantifying each isomer, so three categories were created, “Total CQA”, “Total FQA”, and “Total di-CQA”, to compare across publications (Anthony, Clifford, and Noirot 1993). Each of these categories includes the sum of each isomer in that class; for example, “Total CQA” is a sum of 5-CQA, 4-CQA, and 3-CQA.  27 unique CGAs have been identified in coffee (Clifford et al. 2003; Clifford 2006). The limited recurrences (fewer than 2 publications) of some species led to their exclusion from data collection. 
 

Chlorogenic acids and organic acid concentrations are presented in separate .csv files. A third separate .csv is included that details the full bibliographic information for the publications that provided the data for this dataset.