Research report
A research report is the sourced material a domain
dossier is synthesized from — generated on a plan and a cadence, one topic per file.
A report carries no confidence tags. Its bracketed markers say who might have
an incentive to shade a line; none of them says anyone checked it. To reach a score, a line has
to be drafted onto a candidate as unverified and pass the Verifier or the
Corroborator, like everything else.
Section clinical-evidence · Version 2026-09-01 · Cadence quarterly · Evidence class 2 published · Sources 6 · Supersedes none
Domain: veterinary-diabetes-glucose-monitoring · Scope: Peer-reviewed literature reachable via this repo's `literature` /
Sourcing: Rests on one guideline confirmed to exist and read at abstract depth only
The 2018 AAHA Diabetes Management Guidelines for Dogs and Cats is indexed in PubMed and its abstract was retrieved directly; it frames home blood-glucose monitoring as "ideal but not mandatory" and centers its recommendations on insulin-formulation choice and avoiding hypoglycemia, not on any specific monitoring device — its full PDF returned HTTP 403 to this pass's fetch, the same failure the domain dossier already recorded twice [1]. A reproducibility study of the pre-device standard of care itself — the serial blood glucose curve — found that paired curves from the same cat, generated under identical conditions days apart, led to the same theoretical insulin-dose recommendation only 6 of 14 times at home and only 14 of 28 times comparing a home curve to a clinic curve [2]. Fructosamine, the most widely used non-curve surrogate for glycemic control, did not correlate significantly with continuous-glucose-monitor (CGM) mean glucose in a 24-dog cohort using CGM as the reference standard [3]. An additional client-owned-cat accuracy cohort for the FreeStyle Libre flash system (n=41, not previously in this domain's record) found good correlation but incomplete agreement against ISO 15197:2013 criteria [4], and a foundational validation study of the portable glucometers used as the ground truth across nearly every accuracy study in this domain found that even they do not show statistical agreement with a laboratory reference method, with one of the two meters tested showing a hematocrit-dependent bias [5]. A CGM day/night profile study recorded low-glucose readings in both healthy and diabetic cats at unpredictable times, reinforcing the case for continuous over point sampling while also reporting that most diabetic cats in the study removed their sensors before the study period ended [6].
_Baseline (v1). No prior version; this establishes the starting point for future diffs._
PubMed indexes "2018 AAHA Diabetes Management Guidelines for Dogs and Cats" as a practice guideline (PMID 29314873), and its abstract was retrieved via efetch: "An essential aspect of successful DM management is to ensure that the owner of a diabetic dog or cat is capable of administering insulin, recognizing the clinical signs of inadequately managed DM, and monitoring blood glucose levels at home, although this is ideal but not mandatory" [1] [single-source]. The abstract's stated scope is insulin-formulation selection, avoiding hypoglycemia, and identifying at-risk patients — it does not name a continuous or flash glucose monitor, consistent with a 2018 publication date that predates most of the veterinary CGM/FGM literature in §4 below. This pass repeated the domain dossier's own attempt to fetch the guideline's full PDF directly from aaha.org and received the same HTTP 403 result, confirming rather than newly discovering the block [1] [paywalled]. Whether a later AAHA revision addresses CGM/FGM technology by name remains unread, since only this 2018 record (last indexed guideline of its kind found by PubMed query) was located.
Every accuracy study in this domain compares a monitoring device against the serial blood glucose curve or a portable meter reading, treating that comparator as ground truth. A prospective study of 7 diabetic cats found that was not a safe assumption: two blood-glucose curves generated at home on separate days, with insulin dose, food type and food amount held constant, produced "the same theoretical recommendation for adjustment of insulin dose on 6 of 14 occasions," and comparing a home curve to a same-week clinic curve agreed on only "14 of 28 occasions" [2] [single-source]. The same abstract reports that cats with good glycemic control had more reproducible curves than cats with poor control (4 of 6 good-control pairs agreed, versus 2 of 8 poor-control pairs) [2] [single-source] — meaning curve-based dosing decisions are least reliable in exactly the animals where getting the dose right matters most.
A retrospective study of 24 dogs with treated diabetes mellitus, each wearing a CGM for a mean of 13.1 days and having fructosamine measured shortly after sensor removal, found that "correlations of individual fructosamine concentrations with mean glucose as well as percentage of measurements
15 mmol/L were not significant (p = 0.372, p = 0.129)," and fructosamine "did not differ between
dogs with and without hypoglycaemic episodes (p = 0.64)." ROC analysis against CGM-defined good control gave an AUC of 0.71 (moderate accuracy) and against poor control an AUC of 0.70 that the authors themselves describe as "non-discriminatory"; the positive likelihood ratio at the optimal cutoffs was only 3 [3] [single-source]. The study's own conclusion: fructosamine "is an imperfect surrogate marker for classifying glycaemic control in diabetic dogs and can only complement serial glucose measurements" [3] [single-source] — a directly negative finding about a measure this domain's clinical practice still relies on when a curve or CGM is not running.
A client-owned-cat cohort (n=41) not previously recorded in this domain evaluated the FreeStyle Libre against a portable meter (AlphaTRAK) reference: correlation was good (ρ=0.88, P<0.0001), 53% of interstitial readings were within 15% of the blood reading, and 92.7% fell in the two safest zones of a surveillance error grid, but the device "did not completely fulfil ISO requirements," concluding it is "sufficiently accurate" despite that gap [4] [single-source]. That comparator meter is itself imperfect: a foundational validation study measured two human portable glucometers against a laboratory chemistry analyzer in 69 dogs and 26 cats and found that, despite high simple correlation, "Passing-Bablok regression and Bland-Altman analyses revealed that the data from both (meters) did not show statistical agreement with the reference values," with one meter's readings significantly affected by hematocrit [5] [single-source]. The same study reports that error-grid analysis judged all measurements from both meters would still lead to "acceptable treatment decisions" [5] [single-source] — the same accuracy-vs-clinical-consequence gap seen throughout §4 of the domain dossier, here located one layer further back, in the comparator rather than the candidate device.
A CGM profile study in 6 healthy and 10 diabetic cats (7-day sensor, 24-hour profiles) found healthy cats had higher nocturnal glucose variability than daytime (SD 0.6 vs 0.4 mmol/L, P=0.037; CV 13% vs 10%, P=0.041), while diabetic cats showed no day/night difference in mean, minimum, maximum, SD or CV; diabetic cats with moderate-to-poor control had higher mean and maximum glucose over 24 hours and higher daytime SD than well-controlled cats. Hypoglycemia (<3.5 mmol/L) occurred in 5 of 6 healthy and 4 of 10 diabetic cats, "either during the day or night" [6] [single-source]. The same study reports that "most diabetic cats removed sensors prematurely," though 24-hour curves were still obtained in all animals and "the only adverse effect of the device was mild local skin irritation" [6] [single-source] — a smaller, milder complication signal than the §5 dossier record for FreeStyle Libre in dogs and cats, but from a much smaller cohort (n=10 diabetic cats) than those studies.
[1] "2018 AAHA Diabetes Management Guidelines for Dogs and Cats" — Journal of the American Animal Hospital Association (published 2018, month recorded ambiguously as "Jan/Feb" in the PubMed record, no day given; accessed 2026-09-01). PMID 29314873 — https://pubmed.ncbi.nlm.nih.gov/29314873/ [guideline] [2] "Day-to-day variability of blood glucose concentration curves generated at home in cats with diabetes mellitus" — Journal of the American Veterinary Medical Association (published 2007-04-01; accessed 2026-09-01). PMID 17397339 — https://pubmed.ncbi.nlm.nih.gov/17397339/ [peer-reviewed] [3] "Evaluation of fructosamine concentration as an index marker for glycaemic control in diabetic dogs" — The Veterinary Record (published 2022, month/day not given in the PubMed record; accessed 2026-09-01). PMID 33870504 — https://pubmed.ncbi.nlm.nih.gov/33870504/ [peer-reviewed] [4] "Evaluation of the FreeStyle Libre, a flash glucose monitoring system, in client-owned cats with diabetes mellitus" — Journal of Feline Medicine and Surgery (published 2022-08-01, day not given in the PubMed record; accessed 2026-09-01). PMID 35762266 — https://pubmed.ncbi.nlm.nih.gov/35762266/ [peer-reviewed] [5] "The clinical utility of two human portable blood glucose meters in canine and feline practice" — Veterinary Clinical Pathology (published 2014-03-01, day not given in the PubMed record; accessed 2026-09-01). PMID 24446729 — https://pubmed.ncbi.nlm.nih.gov/24446729/ [peer-reviewed] [6] "Comparison of continuous glucose monitoring profiles during the day and night in healthy and diabetic cats" — Journal of Feline Medicine and Surgery (published 2025-11-01, day not given in the PubMed record; accessed 2026-09-01). PMID 41204810 — https://pubmed.ncbi.nlm.nih.gov/41204810/ [peer-reviewed]
Well established: Nothing in this report rests on two independent, non-affiliated sources — every finding above is a single study or a single guideline abstract, and is marked accordingly. The domain dossier's own §4 accuracy table remains the place where this domain's most multiply-sourced pattern lives (multiple FreeStyle Libre cohorts agreeing on the general zone-A/B clinical-accuracy result); this report deliberately did not re-derive that pattern and instead added cohorts and comparator questions the dossier had not yet reached.
Thin: All six numbered findings are [single-source]: the AAHA guideline's stated scope and its continued full-text block [1]; the day-to-day curve-reproducibility figures [2]; the fructosamine-vs-CGM correlation result [3]; the additional FreeStyle Libre cat cohort [4]; the portable-glucometer validation study [5]; and the CGM day/night profile study [6]. No second, independent study repeating any of these six specific questions was found by the queries run this pass (see below).
Rescoped from class 3: None in this report. Every proposition here is a guideline's stated text or a measured study finding — none asks what a person would do.
Out of scope: CPT descriptors (AMA-licensed, unaffected by this report). Veterinary payer/HCPCS/fee-schedule data — already established at the domain-dossier level as structurally unreachable, not re-attempted here since this report's section is clinical evidence, not reimbursement. MAUDE narratives (no connector exists in this repo). Non-US guideline bodies (e.g. any UK, EU or Australian small-animal diabetes guideline) were not searched for in this pass; only AAHA, the guideline body the domain dossier already treats as the relevant one for a US-market candidate, was pursued.
Not searched vs. not found: A second search for a peer-reviewed source corroborating the fructosamine-vs-CGM correlation finding [3] was run and returned nothing — searched and not found, not merely unsearched. Targeted queries for a randomized or controlled trial comparing CGM-guided versus curve-guided insulin titration on a clinical outcome (hospitalization, remission rate, hypoglycemic-event rate) returned no PubMed records under six distinct phrasings tried this pass — a repeated negative result, consistent with (not contradicting) the domain dossier's own observation that no such head-to-head outcome trial has been found. A single ClinicalTrials.gov query (condition="diabetes mellitus", intervention="glucose monitoring") was run to check whether that registry carries any veterinary-diabetes record; every one of the 20 records inspected was a human trial, confirming rather than newly discovering the registry's human-only structural scope already noted in the domain dossier — this check is not numbered as a source above because it produced a structural confirmation, not a citable finding. A more recent (post-
found in PubMed; whether one exists outside PubMed's index was not established either way.
[inference] The pattern across [2], [3] and [5] — that the pre-device comparator itself (serial curves, fructosamine, and even the portable glucometer used as "ground truth") each carry a measured reliability problem of their own — suggests that some of the accuracy gap reported against CGM/FGM devices elsewhere in this domain's record may reflect an imperfect comparator rather than only imperfect device measurement. No source retrieved in this report or the domain dossier states that reading directly; it is this report's own synthesis of three separately reported findings, and is not a finding to be read into any of [2], [3] or [5] individually.
| Proposition | Evidence class | Resolvable identifier | Dossier section |
|---|---|---|---|
| The 2018 AAHA Diabetes Management Guidelines for Dogs and Cats describes home blood-glucose monitoring as "ideal but not mandatory" and does not name a continuous or flash glucose monitor | 2 — published finding | PMID 29314873 | 4. Standard of care & clinical evidence |
| Paired same-cat blood glucose curves generated at home on separate days led to the same theoretical insulin-dose recommendation in only 6 of 14 occasions, and home-vs-clinic curve pairs agreed in only 14 of 28 occasions, in a 7-cat study | 2 — published finding | PMID 17397339 | 4. Standard of care & clinical evidence |
| Fructosamine concentration did not correlate significantly with CGM-measured mean glucose (p=0.372) in a 24-dog retrospective cohort, and its ROC AUC for detecting poor glycemic control (0.70) was described by the study authors as non-discriminatory | 2 — published finding | PMID 33870504 | 4. Standard of care & clinical evidence |
| In a 41-cat cohort, FreeStyle Libre interstitial glucose readings correlated well with blood glucose (ρ=0.88) but did not completely fulfil ISO 15197:2013 accuracy criteria | 2 — published finding | PMID 35762266 | 4. Standard of care & clinical evidence |
| Two human portable blood glucose meters tested against a laboratory reference analyzer in 69 dogs and 26 cats showed high simple correlation but did not show statistical agreement by Passing-Bablok and Bland-Altman analysis, with one meter's accuracy significantly affected by hematocrit | 2 — published finding | PMID 24446729 | 4. Standard of care & clinical evidence |
| In a 16-cat CGM profile study, hypoglycemia (<3.5 mmol/L) occurred in 5 of 6 healthy cats and 4 of 10 diabetic cats at unpredictable times of day or night | 2 — published finding | PMID 41204810 | 4. Standard of care & clinical evidence |