Study summary · Dogs and cats · Nutrient adequacy · Pedrinelli et al., 2019
Testing the prepared food exposed gaps that ingredients alone could not show.
None of 100 internet recipes met all the protein, fat and mineral recommendations evaluated. Some metal concentrations also exceeded the study’s comparison thresholds, but no animals were fed the diets and clinical harm was not measured.
Why this study matters
Calculated recipes and laboratory-tested meals provide different kinds of evidence. This paper examined food prepared from the instructions, so it helps assess what those recipes actually supplied. It also included cats, whose nutritional requirements cannot simply be inferred from dog recipes.
How the study worked
Samples
100 Portuguese-language recipes from 35 internet sources, intended for healthy adult pets: 75 for dogs and 25 for cats. Recipes presented as snacks, milk replacers or unsuitable for daily feeding were excluded, as were recipes missing ingredient quantities.
What happened?
Researchers purchased ingredients in São Paulo and prepared 500 g samples following recipe instructions. Five cat recipes and two dog recipes contained raw animal products; the remaining recipes with animal products used cooked ingredients. Where a dog recipe requested a vitamin-mineral supplement without details, a commercial supplement was selected and added at its labelled amount.
The comparison
Measured nutrient concentrations were compared with NRC and FEDIAF recommendations using the authors’ assumptions about energy intake. Metal concentrations were compared with the tolerable levels selected by the study, some of which were extrapolated from other mammals. These were benchmark comparisons, not a trial against commercial food.
What was measured?
Protein, fat, fibre, selected minerals and metal concentrations. Analyses included chemical methods for macronutrients and ICP-OES for minerals and metals. Iodine and chloride were not measured; selenium was below detection in 98 samples and was excluded from the adequacy comparison.
What the researchers found
| Outcome | Reported result | How to read it |
|---|---|---|
| Overall nutrient adequacy | 0/100 recipes met all recommendations evaluated. | This concerns measured nutrients; it is not a complete evaluation of every essential nutrient. |
| Multiple shortfalls | NRC: 63/75 dog recipes and 25/25 cat recipes had at least three. FEDIAF: all 100 did. | The benchmark and energy assumptions affect classification. |
| Frequent mineral gaps | Calcium and potassium in dog recipes; iron and zinc in cat recipes. | The study measured food composition, not clinical deficiency. |
| Metal concentrations | Lead, cobalt, mercury, uranium and vanadium exceeded selected thresholds in some samples. | Exceeding a threshold does not demonstrate poisoning; exposure, absorption and the threshold’s relevance matter. |
What strengthens the evidence?
The investigators prepared the foods and measured their composition rather than relying entirely on nutrient databases. They described recipe selection, preparation and analytical methods, included both species and compared results with two nutrient frameworks. Replicate laboratory analyses help assess analytical consistency, although they do not substitute for testing repeated shopping and preparation batches.
What limits the conclusions?
The sample represents selected Portuguese-language internet recipes and ingredients purchased locally, not all home-prepared food. Only seven recipes contained raw animal products, so it cannot establish a reliable raw-versus-cooked comparison. Selecting a supplement for vague instructions introduced an assumption. Vitamins, individual amino acids and several other nutrients were outside the reported adequacy assessment. No feeding, absorption or health outcomes were measured. Cooking water was not tested for metals, and some tolerable levels rely on extrapolation from other species. Correlations with ingredients do not establish which ingredient caused contamination.
What the researchers concluded
The authors concluded that inadequately formulated home-prepared diets may present nutritional risks. They also called for better understanding of chronic metal exposure before drawing firm conclusions about its health impact.
Evidence.pet interpretation
The strongest practical message is about checking the complete formulation. Fresh ingredients, a variety of foods or the presence of meat cannot by themselves demonstrate that a meal meets all requirements. The raw subset was too small to support a broad judgement about raw feeding, and the study did not test professionally formulated personalised recipes as a separate category. Treat the metal findings as a reason for further investigation, not a diagnosis of toxicity or a universal claim about particular vegetables or meats. Ingredient origin and preparation can matter, while a laboratory concentration alone cannot tell us what a pet absorbs. For a home-prepared feeding plan, evidence of nutritional adequacy should cover the exact recipe, its supplements and its intended species and life stage.
Funding and author interests
The authors declared no competing interests. The publisher text reviewed did not identify a separate funding statement. Laboratory work included facilities at the University of São Paulo and Biorigin Brazil; that acknowledgement of facilities does not by itself establish who funded the study.
Read the original research
Pedrinelli V, Zafalon RVA, Rodrigues RBA, et al. Concentrations of macronutrients, minerals and heavy metals in home-prepared diets for adult dogs and cats. Scientific Reports. 2019;9:13058. doi:10.1038/s41598-019-49087-z.
Evidence.pet editorial summary · Study ID STU-014 · Source checked 22 September 2026. Strengths, limitations and interpretation are our editorial assessment, not a formal evidence grade.
Adapted from Pedrinelli and colleagues (2019), Scientific Reports, under CC BY 4.0. Evidence.pet condensed the findings and added independent interpretation. Publisher full text checked; supplementary datasets were not independently reanalysed.
Creative Commons Attribution 4.0 licence. The source has been condensed and reorganised.