The Best Over-the-Counter Continuous Glucose Monitors of 2026, Ranked by Accuracy, FDA Clearance and What You Actually See

How we score: the Dr. Axe Clean Glucose Score™

Most CGM reviews rank on app features and how the tester felt. Ours starts with what you can check: who tested the accuracy and on whom, what FDA's own records say, what the app really shows, what the privacy policy allows, and whether the marketing stays inside the label. It's an editorial assessment built from FDA databases, user guides, privacy policies, product pages and studies, not our own testing. Price isn't scored; see the cost chart.

One caveat applies to every score: only one sensor has been tested independently in the people who buy these most, and in just 39 young adults. Every other accuracy number comes from the makers, in people with diabetes.

Why we don't chase every spike

That's our dividing line: we reward monitors that help you see patterns and sell themselves honestly, and we mark down marketing that turns a normal post-meal rise into a disease warning or a weight-loss promise.

What to do: judge patterns over days, not single peaks. Look at how high a meal takes you above your starting point and how long it takes to come back. If your numbers worry you, get a lab fasting glucose or A1C[64].

Does wearing a CGM help? What the evidence shows

The solid findings describe what glucose does in healthy people. The findings on whether wearing a CGM changes your health are thin: a 2026 systematic review of 12 studies rated the dietary evidence low certainty and the rest very low[58]. Here's how we'd grade it:

Two practical takeaways: use a CGM as a two-week teacher, not a lifelong scoreboard, and trust the pattern over the peak.

How to read a glucose monitor's specs

The same 8.3% can mean very different things depending on who wore the sensor and what it was compared with[23][1], and the update interval and display range decide what you'll actually see[3][8]. Here's what each spec tells you:

More picks by category

Best ways to use a Dexcom Stelo

Five ways to buy the same 15-day sensor. The cheapest is Stelo direct; the programs add coaching, labs or a different app, and all run through a Dexcom account[43][51].

Best ways to use an Abbott Lingo

Minute-by-minute readings and the only independent accuracy data in healthy adults[1]. Ultrahuman ships Lingo in its original packaging[33].

Best with a human coach

Nutrisense recommends two dietitian calls a month, billed to insurance[45]; Signos offers in-app dietitian chat[41]; Ultrahuman includes performance coaches[33]. Stelo and Lingo both point you to insurance-billed dietitians too[22][31].

Best under $100 a month

Our per-30-day costs; Oura assumes you already own the ring.

Smart rings and watches that claim glucose

In February 2024 FDA told people not to use smartwatches or smart rings that claim to measure blood glucose “without piercing the skin,” because inaccurate readings can lead to wrong decisions; the warning covers any brand[19]. That's different from a watch or ring app that shows data from a real CGM: Stelo can send readings straight to an Apple Watch[25], and the Oura App shows Stelo data[36]. If a wearable promises glucose with no sensor under the skin, skip it.

Prescription CGMs and devices we couldn't rate

These are diabetes devices with alarms and insulin-pump links, sold by prescription. We list them for context, not as recommendations for people without diabetes.

What about the Dexcom warning letter and the Libre 3 recall?

Both sensor makers have had a warning letter and recalls of their prescription products: Abbott's Libre 3 correction, and Dexcom's G7 app, receiver and diverted-sensor recalls[13]. None involves Stelo or Lingo. Here's what the records say.

What we'd skip

Two things to avoid, whichever monitor you buy.

When the label says more than the test

The biggest gaps aren't lies; they're fine print. About 1 in 5 sensors doesn't last its full wear time, accuracy slips on day one and after sugar, and Stelo's newest clearance hasn't reached its app or site yet.

Frequently asked questions

References

1. Independent test Prakash N, et al. Agreement of over-the-counter continuous glucose monitors with capillary-meter glucose in young adults without diabetes. J Diabetes Sci Technol, 2026 (39 adults without diabetes, University of South Florida; Abbott Lingo biosensor vs Contour Next fingerstick meter; overall MARD 12.9%, bias −4.8 mg/dL, 66.5% within 15%/15 mg/dL; MARD 15.3% after a sugar-sweetened drink vs 10.6% after placebo; 6.2-minute estimated lag, MARD 11.2% with a 5-minute shift; 13.8% of trend arrows moderate-to-extreme risk; “No independent study has yet evaluated OTC CGM accuracy”; no direct funding). Cited as independent because no maker funded it. Open source
2. FDA database FDA 510(k) database via openFDA (data current to Sept. 21, 2026), searched by DrAxe.com on Sept. 30, 2026: every clearance under the two over-the-counter CGM product codes (SAF, non-intensive monitoring; SBH, non-intensive management): K234070, K261359, K260935 and K262472 (Dexcom Stelo), K233655 (Abbott Lingo), K233861 (Abbott Libre Rio), K250106 (Signos). No other company holds one. Open source
3. FDA database FDA 510(k) K234070, Dexcom Stelo Glucose Biosensor System, cleared Mar. 5, 2024 (code SAF): adults 18 and older not on insulin; “The Stelo System uses the same hardware design as the predicate G7 CGM System”; app displays glucose every 15 minutes (sensor sends every 5); 70–250 mg/dL; “does not provide any glucose or system alerts.” Open source
4. FDA database FDA decision summary for K234070: MARD by wear period 9.1% (beginning), 7.8%, 7.3%, 9.1% (end) in 130 adults; 19.8% of sensors (26 of 131) had early sensor shut-off; 73.5% of evaluable sensors lasted to the final day in the original study. Open source
5. FDA database FDA 510(k) K261359, Stelo (special 510(k)), cleared May 21, 2026: app changes including glucose response scores; same 18+ indication; 70–250 mg/dL; no glucose alerts. Open source
6. FDA database FDA 510(k) K260935, Stelo, cleared June 12, 2026 (code SBH): “people 2 years and older not on insulin” (caregiver supervision under 18); reportable range 55–350 mg/dL; direct-to-Apple-Watch; “not for people on dialysis, with problematic hypoglycemia, are modifying medication without HCP consultation, or have a history of eating disorders”; no glucose alerts; pediatric accuracy on days 11–15 “not directly assessed.” Open source
7. FDA database FDA 510(k) K262472, Stelo alternate adhesive patch, cleared Aug. 11, 2026; same 2+ indication as K260935. Open source
8. FDA database FDA 510(k) K233655, Abbott Lingo Glucose System, cleared May 29, 2024 (code SAF): 18 and older not on insulin; based on the FreeStyle Libre 2 sensor (K222447); readings every 1 minute; 55–200 mg/dL; “There are no alarms”; 1-hour warm-up. No new clinical study: Abbott leveraged FreeStyle Libre 2 clinical data from people with diabetes as “a worst-case scenario” and states the system meets the iCGM special controls. Open source
9. FDA database FDA 510(k) K233861, Abbott Libre Rio CGM System, cleared June 7, 2024 (code SBH): non-insulin-using persons 18 and older; 40–400 mg/dL; up to 15 days; no glucose alarms. Open source
10. FDA database FDA 510(k) K250106, Signos Glucose Monitoring System (mobile app), cleared Mar. 21, 2025 (code SAF): displays data from the Stelo sensor; 18 and older not on insulin; “This information may be useful in helping users to maintain a healthy weight”; 70–250 mg/dL every 15 minutes; glucose alerts: none. Open source
11. FDA database FDA 510(k) K243214, Dexcom G7 15 Day Continuous Glucose Monitoring System (prescription, code QBJ), cleared Apr. 9, 2025. Open source
12. FDA database FDA premarket approval database via openFDA: Eversense (Senseonics) P160048, original approval June 21, 2018, latest approved change (S031) Mar. 19, 2026; Medtronic P160007 (Guardian Connect, 2018) and P250012 (Simplera Sync, approved Aug. 6, 2025; an approved change covering Simplera and Simplera Sync May 26, 2026). Open source
13. FDA database FDA device recall database (searched via openFDA, Sept. 30, 2026; link is FDA’s public recall search): no recall of Stelo, Lingo, Libre Rio or Signos. Dexcom recalls since 2025 include the G7 receiver speaker (Z-1944-2025, 602,445 units), G7/ONE+ app software with no “Sensor Failed” alert (Z-2446 to Z-2450-2025), a G7 app defect (Z-2163-2026) and stolen scrap G7 sensors resold (Z-2574-2026). FreeStyle Libre 3 and Libre 3 Plus correction initiated Nov. 24, 2025 (Z-1020-2026 to Z-1023-2026; 1,060,723 + 1,663,241 + 258,913 + 8,062 sensors); Medtronic Simplera sensor recall Z-0033-2026 (cap label may not adhere, sterility). Open source
14. FDA database FDA, posted company announcement: “Abbott Initiates Medical Device Correction for Certain FreeStyle Libre 3 and FreeStyle Libre 3 Plus Sensors” (Nov. 24, 2025): incorrect low glucose readings; 736 severe adverse events (57 in the U.S.) and 7 deaths (none in the U.S.) reported globally; “no other Libre products (FreeStyle Libre 14 day, FreeStyle Libre 2, FreeStyle Libre 2 Plus, or Libre Pro sensors) or Abbott biowearables are impacted.” Open source
15. FDA database FDA warning letter to Dexcom, Inc., CMS 700835, Mar. 4, 2025: after inspections in San Diego (Oct.–Nov. 2024) and Mesa, Ariz. (June 2024), FDA cited G6 and G7 sensors as adulterated and misbranded; Dexcom changed “a component used in the resistance layer” of its sensors; its equivalence study “resulted in determining your primary endpoints were not met”; sensors with the new component showed “significantly greater variability.” No close-out letter was posted as of Sept. 30, 2026. Open source
16. FDA database FDA warning letter to Abbott Diabetes Care, Inc., CMS 722046, Jan. 23, 2026 (posted Feb. 3, 2026): after an October 2025 inspection in Alameda, Calif., which makes “FreeStyle Libre Flash and Continuous Glucose Monitoring Systems including the FreeStyle Libre 3,” FDA said Abbott “releases finished FreeStyle Libre integrated continuous glucose monitoring system (iCGM) devices without performance-testing the finished ... devices for accuracy prior to distribution” and found its responses inadequate. Open source
17. Public record / study MedTech Dive, Feb. 3, 2026: “Abbott receives FDA warning letter over FreeStyle Libre CGMs”; analysts wrote the letter does not require a product recall and that Abbott “is working to implement corrective actions.” Open source
18. FDA database FDA MAUDE adverse-event reports via openFDA (data current to Sept. 22, 2026), counted by DrAxe.com: Stelo, Sept. 9, 2024 to Aug. 31, 2026: 2,448 malfunction, 156 injury, 7 other; Lingo, Oct. 17, 2024 to Aug. 25, 2026: 69 injury, 33 malfunction. Reports are unverified, depend on how many people use a device and how makers report, and are not rates. Open source
19. FDA database FDA Safety Communication: “Do Not Use Smartwatches or Smart Rings to Measure Blood Glucose Levels” (Feb. 21, 2024): applies to devices that claim to measure blood glucose “without piercing the skin,” unlike apps that display data from FDA-authorized CGMs. Open source
20. Public record / study 21 CFR 862.1355, integrated continuous glucose monitoring system special controls: e.g., within ±20% of the lab reference across the range more than 87% of the time (lower confidence bound), within ±15% at 70–180 mg/dL more than 70%, clinical data in the intended-use population. Open source
21. Maker's data Stelo by Dexcom homepage, Sept. 30, 2026: benefit list “Stabilize energy,” “Improve focus,” “Reveal patterns around mood,” “Improve metabolism,” “Lower prediabetes risk,” “Slow diabetes progression”; footnote “Results obtained from previous Dexcom device(s) in patients with diabetes”; paid spokesperson labeled. Open source
22. Maker's data Stelo buy page, Sept. 30, 2026: $99 one-time (2 biosensors); $89 1-month subscription ($99 struck, “SAVE 10%”); $252 3-month subscription ($297 struck), “starting at just $84/month”; “For people not using insulin, ages 18 years and older”; don’t use with a history of eating disorders; 15-day lifespan + 12-hour buffer; 30-minute warmup; HSA/FSA eligible; “All sales are final”; 77.9% of sensors lasted 15 days. Open source
23. Maker's data Stelo User Guide AW-1000421-10 Rev 004 (rev. date 07/2026): 130 adults with type 1 or type 2 diabetes at 6 U.S. centers, 20,330 paired readings vs YSI lab analyzer: 94.0% within 20/20%, 88.6% on Day 1, MARD 8.3%; readings shown only between 70 and 250 mg/dL; sends data to Apple Health or Health Connect “with a 3-hour delay”; non-diabetes target range 70–140 mg/dL; 77.9% of sensors lasted 15 days; 18 and older. Open source
24. Maker's data Stelo blog, “Stelo accuracy”: “Stelo is 93% accurate” meaning “93% of Stelo readings fall within 20/20% YSI” lab criteria. Open source
25. Maker's data Stelo app compatibility page: minimum iOS 18.6 or Android 13; app 3.3 and later sends glucose directly to Apple Watch (watchOS 11.6). Open source
26. Maker's data Dexcom Privacy Notice (last updated July 2024), on stelo.com: disclosure of personal data to “advertising networks” and “social networks” “qualifies as the “sale” of Personal Data” or targeted advertising under certain state laws; opt-out offered; “we do not “sell” any consumer data in the traditional sense for monetary compensation”; rights to deletion and portability. Open source
27. Public record / study MedTech Dive, Mar. 26, 2025: “Dexcom warning letter...”: Dexcom’s COO said there was “no need yet to recall products”; an analyst wrote Dexcom stopped in-sourcing the compound and reverted to the external supplier. Open source
28. Public record / study MedTech Dive, Mar. 27, 2025: Dexcom spokesperson: “Dexcom qualified a second source for one of its raw materials,” “extensive testing” showed it met specifications, no design changes were made, it notified FDA “through 510(k) submissions, following the guidelines for non-significant changes,” and it is “committed to resolving this issue promptly.” Open source
29. Public record / study Levens v. Dexcom, Inc., No. 3:25-cv-02565 (S.D. Cal., filed Sept. 29, 2025; amended Nov. 4, 2025; open on CourtListener, Sept. 30, 2026); a related case, Dalora v. Dexcom, Inc., No. 3:25-cv-03210, was noticed Nov. 25, 2025. ClassAction.org summary: the proposed class action alleges the G6 and G7 material change caused “considerably greater variability.” Allegations only. Open source
30. Maker's data Lingo homepage, Sept. 30, 2026: menu items “Lower prediabetes risk,” “Manage your weight,” “Support heart health”; “Tried everything, but still not losing weight?”; “A CGM can help provide insights that support weight-loss efforts”; sign-up promising insights “to supercharge your metabolism.” Open source
31. Maker's data Lingo plans page, Sept. 30, 2026: $54 2-week plan (1 biosensor), $89 4-week plan (2), $249 12-week subscription (6); HSA/FSA eligible; 30-day money-back on first purchase; iOS and Android; data appears in 60 minutes; 77.1% of biosensors lasted 14 days, 14.7% may last under 11; Apple Health and Health Connect integration; 18 and older not on insulin. Open source
32. Maker's data Lingo Privacy Notice (Global), July 2026: “We do not sell your Personal Information for money, but some of the ways in which we may disclose your Personal Information to certain third parties ... such as analytics and targeted advertising services, may be considered a “sale””; categories: identifiers, internet activity, geolocation; account deletion by request, including in the app. Open source
33. Maker's data Ultrahuman M2 Live pricing page (U.S.), Sept. 30, 2026: “No prescription needed”; subscription $99 for “2 CGM Sensors/Month,” “At Introductory Pricing,” “save 23%”; “In the US, we integrate with Abbott’s Libre 3 Plus and Lingo Biosensors”; “We use ... Abbott Lingo™, in their original packaging”; NFC needed to activate; 14-day sensors; performance coaches included; “Lose weight sustainably”; “reducing the risk of T2 diabetes, obesity, PCOS/PMOS, hypertension”; “improve cognitive performance.” Open source
34. Maker's data Ultrahuman Privacy Policy (updated Sept. 13, 2026): “We do not sell your personal information”; “We use pixel tags to collect personal information from website visitors, which may then be used to ... provide tailored advertisements”; third parties include Facebook Ads and Google Ads; rights to deletion and a portable copy. Open source
35. Maker's data Oura store, Stelo Glucose Biosensor page: pair Stelo with an Oura Membership to see glucose in the Oura App; 2-pack = 30-day supply; “For people NOT on insulin, ages 18 years or older”; “No prescription needed and FDA-cleared”; HSA/FSA eligible. Open source
36. Maker's data Oura Member Care, “Glucose Tracking” (July 2, 2026) and “Stelo by Dexcom Integration” (July 6, 2026): “you’ll need both an Oura Membership and a Stelo account”; U.S. only; Gen3 or later ring; “We currently integrate only with the Stelo biosensor”; only glucose data flows from Stelo to Oura; meal-level time above range. Open source
37. Maker's data Oura Membership page: $5.99 a month; first month free for new members; HSA/FSA eligible. Open source
38. Maker's data Oura “Metabolic Health” page: “so you can feel full of energy every day”; “Beat cravings and energy slumps”; tips toward “more sustained energy levels, better sleep, a healthy weight.” Open source
39. Maker's data Oura Privacy Policy: “We do not sell or share your personal data for cross-context behavioral advertising”; website cookies used “to create audiences for online advertisement”; Membership Hub data export; right of deletion. Open source
40. Public record / study Fierce Healthcare, Nov. 19, 2024: Dexcom invested $75 million in Oura’s Series D, valuing Oura at more than $5 billion, and the companies agreed to integrate Stelo data with the Oura Ring. Open source
41. Maker's data Signos plans page, Sept. 30, 2026: “Save 25% with code AUTUMN25”; 6-month CGM plan $127/month ($169 regular), 3-month $150 ($199), 1-month $225 ($299); 2 sensors a month; “All plans include the Stelo glucose biosensor by Dexcom”; “Hit Your Weight Goals,” “Discover Your “Fat-Loss Foods”,” “Supercharge Your Metabolism”; “All sales final. No refunds or exchanges”; HSA/FSA “Possibly.” Open source
42. Maker's data Signos Privacy Policy: personal information used to provide “advertising and promotional information (e.g., targeted ads, retargeted ads)”; shared with service providers including “advertising partners”; right to delete. Open source
43. Maker's data Signos Help Center, “The Dexcom Authorization Process [Stelo]” (updated Feb. 5, 2026): a one-time Dexcom account and authorization step when setting up Stelo. Open source
44. Maker's data Nutrisense homepage: “24/7 glucose data and expert coaching to help you cut cravings, boost energy, and lose weight, for good”; “Understand your body’s data and what’s behind your exhaustion, weight, and anxiety.” Open source
45. Maker's data Nutrisense products page, Sept. 30, 2026: “15% Off All Programs”; 6-month $179 → $152/month, 3-month $209 → $178, month-to-month $249 → $212; monthly delivery of 2 Stelo biosensors with Stelo app access; 2 dietitian video calls a month recommended, “Most pay $0 out-of-pocket with insurance coverage”; Nutrisense app iOS 16.2+/Android 12+; all sales final. Open source
46. Maker's data Nutrisense Privacy Policy (last updated Apr. 13, 2026): “We do not sell any Personal Information”; “Nutrisense does not track its users over time and across third party websites to provide targeted advertising”; rights to delete and to a portable copy. Open source
47. Maker's data Nutrisense Help Center, “How to Set Up Your Stelo Sensor with the Nutrisense App”: download the Stelo by Dexcom app and scan the applicator’s QR code. Open source
48. Public record / study MedTech Dive, June 5, 2026: “Dexcom has bought Nutrisense,” which sells subscription packages that include Stelo and dietitian coaching; terms not disclosed. Open source
49. Maker's data Levels homepage, Sept. 30, 2026: “Bring your own CGM or get one through Levels”; app $15/month or $80 billed annually; Core $399/year, Complete $1,329/year; “Among members who started in the prediabetes range, most improved their level and many reversed course” (“Based on retrospective analysis of Levels members. Not a clinical trial”); programs for “glucose control, heart health, and weight loss”; “energy, cravings, and focus.” Open source
50. Maker's data Levels Support, “Stelo Biosensor Pricing” (June 18, 2026): $89 per subscription shipment or $99 one-time two-pack; HSA/FSA reimbursement through Truemed. Open source
51. Maker's data Levels Support, “Bring Your Own Sensors to Levels” (June 1, 2026): Stelo connects “by logging into your Dexcom/Stelo account in the Levels app”; Dexcom G7 and G6 also connect; Levels “can connect you with a physician” for a G7 prescription. Open source
52. Maker's data Levels Health Privacy Policy: “We don’t sell (or sell access to) your data to third parties (for advertising, or for other purposes)”; export biometric data and logs “in CSV format” at app.levels.com; deletion by emailing privacy@levels.com. Open source
53. Maker's data Veri homepage, Sept. 30, 2026: “Veri is joining ŌURA”; no pricing page found. Open source
54. Public record / study Shah VN, et al. Continuous glucose monitoring profiles in healthy nondiabetic participants: a multicenter prospective study. J Clin Endocrinol Metab, 2019 (153 people aged 7–80, blinded Dexcom G6 up to 10 days; mean glucose 98–99 mg/dL, 104 over age 60; median 96% of time at 70–140 mg/dL; 2.1% above 140, about 30 minutes a day; 1.1% below 70, about 15 minutes). Open source
55. Public record / study Hall H, et al. Glucotypes reveal new patterns of glucose dysregulation. PLoS Biol, 2018 (57 people without a diabetes diagnosis; 30 ate standardized breakfasts; “cornflakes and milk caused glucose elevation in the prediabetic range (>140 mg/dl) in 80% of individuals”; normoglycemic people reached prediabetic and diabetic ranges 15% and 2% of the time). Open source
56. Public record / study Zeevi D, et al. Personalized nutrition by prediction of glycemic responses. Cell, 2015 (800 people, week-long CGM, 46,898 meals; “high variability in the response to identical meals”; predictions validated in 100 more people). Open source
57. Public record / study Shilo S, et al. Continuous glucose monitoring and intrapersonal variability in fasting glucose. Nat Med, 2024 (8,315 adults without diabetes aged 40–70; 59,565 mornings; mean fasting glucose 96.2 mg/dL; day-to-day SD 7.52 mg/dL; of 5,328 with normal fasting glucose on the first morning, 40% would have been reclassified into the prediabetes range and 3% into the diabetes range on a later morning). Open source
58. Public record / study Makkena HB, et al. Behavioral and psychosocial effects of participant-visible continuous glucose monitoring in adults without diabetes: a systematic review. Medicina, 2026 (12 studies; one 8-week trial of CGM + low-glycemic education vs education alone cut carbohydrate intake 15.2 g/day; physical activity unchanged; psychosocial evidence “insufficient to establish benefit or exclude harm”; certainty low to very low). Open source
59. Public record / study Richardson KM, et al. Understanding the benefits and psychological burdens of using continuous glucose monitoring for lifestyle change. Obes Res Clin Pract, 2025 (56 adults not on insulin; 68% reported fear of type 2 diabetes when they saw high readings; distress linked with younger age and obesity; eating-disorder symptoms correlated with distress about the sensor’s appearance). Open source
60. Public record / study Shukla AP, et al. Food order has a significant impact on postprandial glucose and insulin levels. Diabetes Care, 2015 (11 adults with type 2 diabetes; ciabatta and orange juice, grilled chicken breast, salad and broccoli; vegetables and protein first cut glucose 28.6%, 36.7% and 16.8% at 30, 60 and 120 minutes; 2-hour glucose rise 73% lower). Open source
61. Public record / study Shukla AP, et al. Carbohydrate-last meal pattern lowers postprandial glucose and insulin excursions in type 2 diabetes. BMJ Open Diabetes Res Care, 2017 (16 adults; glucose rise 53% lower and peak 54% lower with carbohydrate last vs first). Open source
62. Public record / study Buffey AJ, et al. The acute effects of interrupting prolonged sitting time with standing and light-intensity walking. Sports Med, 2022 (7 one-day randomized crossover trials, mostly adults with overweight or obesity; walking breaks vs sitting: post-meal glucose effect size −0.72; standing −0.31; walking vs standing −0.30; typical protocol 2 minutes every 20 minutes). Open source
63. Public record / study Shishehbor F, et al. Vinegar consumption can attenuate postprandial glucose and insulin responses; a systematic review and meta-analysis. Diabetes Res Clin Pract, 2017 (glucose AUC standardized mean difference −0.60; insulin −1.30; trials had “conflicting results”). Open source
64. Public record / study CDC, “Diabetes Testing”: fasting blood sugar 99 mg/dL or below normal, 100–125 prediabetes, 126 or above diabetes; 2-hour glucose tolerance 140 or below normal, 140–199 prediabetes; A1C below 5.7% normal. These are lab tests. Open source

How we make money: We may earn a commission when you buy through links on this page. It never changes our rankings: products are scored before any affiliate terms are discussed, brands can't pay for placement, and we recommend products whether or not they have an affiliate program. Scores are editorial assessments based on FDA records, brand documents and studies, not our own lab testing.