What a Single Lab Result Can’t Tell You

Findings from 96,797 patient profiles connect nutrient status with liver, kidney, and metabolic markers, offering a closer look at how lab results fit together.

When a ferritin result comes back higher than expected, what else on the lab report might help explain it? With elevated homocysteine, the first thought may be folate or vitamin B12, but could kidney function be relevant?

These are the kinds of clinical patterns explored in this issue of Clinical Insights Pulse.

In our previous issue, we examined more than 200,000 nutrient biomarker results and assessed how often they fell outside Fullscript’s optimal ranges. In this Insights Pulse, we explored how certain biomarker results might be related.

We analyzed de-identified laboratory data from Fullscript and Rupa, examining 39 biomarker pairings across 96,797 patient profiles. Several familiar physiological relationships were visible, while others were surprisingly weak. Three groups of findings illustrate how selected biomarkers appeared together across tested profiles. 

At a Glance

Using the Fullscript optimal ranges applied in this analysis:

• Higher ferritin was associated with higher ALT. ALT was above optimal in 13.6% of profiles with above-optimal ferritin, compared with 7.5% of those with optimal ferritin. The relationship with the inflammatory marker hsCRP was much weaker.
• Homocysteine reflected more than B-vitamin status. Above-optimal homocysteine was more common with lower folate and B12, but also with lower estimated kidney filtration. Nutrient results alone did not tell the whole story.
• Lower omega-3 status was associated with a less favorable triglyceride pattern. Triglycerides were above optimal in 33.2% of profiles with below-optimal omega-3 status, versus 16.1% at or above optimal omega-3.

How We Looked

We examined laboratory records collected between January 1, 2023 and March 31, 2026. For each biomarker pair, both results came from the same patient profile and sample. We retained only the earliest test result for each pair, so repeated testing did not give some profiles more weight.

First, we assessed whether the numerical results tended to rise or fall together, independently of any laboratory ranges, using Spearman’s rho (ρ), which measures the direction and strength of an association. Values closer to 1 or -1 indicate stronger relationships, while values closer to zero indicate weaker ones. We then grouped selected results using Fullscript’s optimal ranges and compared how often a related marker was above or below optimal. These comparisons are expressed as prevalence ratios: a ratio of 2 means the finding was twice as common in one group as in the comparison group. We also checked whether relationships differed by source, sex, or age where information was available.

A note on ranges: “Below optimal” and “above optimal” refer to the Fullscript optimal ranges used for this analysis. They do not necessarily mean a result was outside the testing laboratory’s reference range. Where the figures use “elevated,” they mean above Fullscript optimal. The cutoffs describe the comparisons, not diagnoses or treatment targets.

Finding 1: Ferritin – Looking Beyond Iron Stores

Ferritin is generally useful because it tells us about iron stores, yet iron is not the only reason it can rise. Inflammation and liver dysfunction can increase ferritin independently of iron overload, which is why an elevated result may be interpreted alongside other relevant clinical information. (9)

The Fullscript findings put numbers around one part of that assessment.

Among 12,957 profiles with both ferritin and ALT results, above-optimal ALT was relatively uncommon overall, but it was nearly twice as prevalent in profiles with above-optimal ferritin (13.6%) compared with those with optimal ferritin (7.5%), representing a prevalence ratio of 1.82. In other words, above-optimal ALT was 82% more prevalent among profiles with above-optimal ferritin. When we looked at the actual numerical values rather than grouping results by optimal range, ferritin and ALT also showed a modest positive association in both females (ρ = 0.215) and males (ρ = 0.183).

The definitions matter here. Above-optimal ferritin began at 75 ng/mL for females and 79 ng/mL for males; above-optimal ALT began at 29 U/L and 46 U/L, respectively. These are Fullscript defined optimal ranges, not thresholds for diagnosing iron overload or liver disease. Some providers and liver guidelines may use lower ALT cutoffs, so more results would be considered elevated under those frameworks. (7) That means the absolute frequency of elevated ALT alongside elevated ferritin could be higher in practices using lower thresholds, although we did not test whether the relative association would also be stronger.

Across the other markers examined, the ferritin relationships varied considerably in strength. The strongest was with TIBC (ρ = -0.460 in females; -0.323 in males). That inverse relationship fits the physiology of iron balance: ferritin tends to fall while transferrin and TIBC rise as iron stores are depleted, while inflammation can produce the opposite pattern by increasing ferritin and lowering transferrin. GGT also showed a modest positive association with ferritin (ρ = 0.266; 0.232). Ferritin can rise in liver disease and obesity, while GGT is also commonly elevated in metabolic dysfunction and fatty liver, although neither is specific to a single cause. (9)(2) AST was less consistent (ρ = 0.164; 0.010) and was influenced substantially by age and sex. Unlike ALT, AST is also found in skeletal and cardiac muscle, which may partly explain why its relationship with ferritin was less clear. (7)

The inflammatory relationship was less pronounced. In the separate ferritin-hsCRP comparison, hsCRP was above optimal in 58.0% versus 53.9% of the higher-ferritin and optimal-ferritin groups. The ferritin-hsCRP association was also weak. This does not contradict ferritin’s role as an acute-phase reactant, but it cautions against solely focusing on ferritin as a substitute for an inflammatory marker. 

Why It Matters

When ferritin does not fit the expected iron picture, it is worth asking whether liver and inflammatory results provide another explanation. World Health Organization (WHO) guidance advises against using ferritin alone to identify iron overload and recommends clinical and laboratory evaluation of the underlying cause. (9) Recent American Society of Hematology guidance makes a complementary point from the iron-deficiency perspective: in adults with anemia of inflammation, ferritin should be interpreted alongside transferrin saturation rather than used alone to assess iron status. (8) In this analysis, the fact that ferritin was associated differently with TIBC, ALT, GGT, AST, and hsCRP reinforces that an elevated ferritin result can sit within very different clinical contexts. 

The practical question is not whether every ferritin result requires additional testing. It is whether the available iron, liver, and inflammatory markers provide enough context to interpret why ferritin is abnormal.

Want help explaining this to your patients?

A patient friendly version can be found in our blog entitled, “Your Ferritin Result Isn't Always About Iron”.

Finding 2: Homocysteine Is Not Just a B-Vitamin Story

Homocysteine is an amino acid produced during methionine metabolism. (3) Its concentration is influenced by B-vitamin-dependent pathways and kidney function, and elevated levels have been associated with cardiovascular disease and dementia risk, making it a useful marker to interpret in clinical context rather than in isolation. (3)(10)

The folate and B12 relationships were clear, but incomplete.

Among 13,914 profiles with both folate and homocysteine measured, homocysteine was above optimal in 48.1% of profiles with below-optimal folate, versus 25.7% with optimal folate. It was nearly twice as common in the lower-folate group, with a prevalence ratio of 1.87. This comparison used folate below 14.8 ng/mL and homocysteine at or above 10 µmol/L. Folate and homocysteine also showed a clear inverse relationship in the association analysis (ρ = -0.347).

Vitamin B12 showed a similar pattern. Across 21,785 profiles with both tests, above-optimal homocysteine was present in 49.7% of those with B12 below 500 pg/mL, compared with 32.5% of those with B12 from 500 to below 800 pg/mL. The prevalence ratio was 1.53.

These observations fit the underlying physiology. Folate and B12 help convert homocysteine back into methionine. When either nutrient is insufficient, or when folate-dependent metabolism is impaired, homocysteine can accumulate. Genetic variation in MTHFR can also influence this pathway, particularly in the context of lower folate status. (5) Serum folate, B12, and homocysteine therefore provide physiologically related, but not interchangeable, information.

Kidney Function Adds Another Piece

The association with renal function (eGFR) makes this more than simply a B-vitamin story.

Among 16,951 profiles with homocysteine and estimated glomerular filtration rate (eGFR) measured together, homocysteine was above optimal in 49.1% of profiles with eGFR below 90 mL/min/1.73 m², versus 26.7% with eGFR at or above 90. The prevalence ratio was 1.84. Homocysteine and eGFR showed a similarly sized inverse relationship (ρ = -0.355).

Reduced kidney function (lower eGFR) can be associated with higher homocysteine, likely in part because reduced filtration is accompanied by changes in homocysteine metabolism and clearance. (4) In this analysis, the relationship weakened when we looked within age and sex groups, suggesting that some of the overall association may reflect differences in the age and sex of the tested population rather than kidney function alone. Practically, eGFR can provide useful context when homocysteine is elevated, but it should not be assumed to be the sole explanation. An eGFR below 90 in this analysis should not be read as a diagnosis of chronic kidney disease. (6)

Why It Matters

An elevated homocysteine result can be tempting to interpret mainly as a folate or B12 problem. These findings suggest the picture may be broader. In this dataset, lower estimated kidney filtration (eGFR) was associated with higher homocysteine to a similar degree as lower folate. Even among profiles with optimal folate, more than a quarter still had above-optimal homocysteine.

For providers, the surrounding results can help explain what may be contributing. Low folate or B12 may support a nutrient-related explanation, while lower eGFR points to kidney function as another important factor. Because creatinine-based eGFR is an estimate and has important limitations, cystatin C may help refine the assessment when kidney function remains uncertain or a more accurate estimate would change clinical decision-making. (6) The broader question is not simply whether folate or B12 should be addressed, but whether nutrient status, kidney function, and the rest of the clinical picture together help explain why homocysteine is elevated.

Want help explaining this to your patients?

A patient friendly version can be found in our blog entitled, “Your Homocysteine Result Isn't Just a B-Vitamin Story”.

Finding 3: Omega-3 Status Belongs in the Metabolic Conversation

A previous insights pulse highlighted how frequently omega-3 results fell below Fullscript’s optimal range among tested profiles. This analysis asked a different question: how did omega-3 status relate to lipid and inflammatory markers measured alongside it?

Here, omega-3 status refers specifically to the reported EPA+DPA+DHA measure.

The strongest omega-3 relationship in this analysis was actually with vitamin D (ρ = 0.347). Both nutrients can share dietary and supplementation sources, and some research suggests omega-3 intake may also influence vitamin D status. (1) However, this analysis cannot distinguish those explanations from other shared patient or testing characteristics, so the relationship should not be interpreted as evidence that one nutrient directly determines the other.

The relationships with triglycerides and hsCRP were weaker, but more directly relevant to the metabolic question explored here. Among 3,372 profiles with EPA+DPA+DHA and triglycerides tested together, triglycerides were above optimal in 33.2% of profiles with below-optimal omega-3 status, compared with 16.1% at or above optimal status. That is roughly one in three versus one in six, with a prevalence ratio of 2.05. The association between EPA+DPA+DHA and triglycerides was more modest (ρ = -0.178).

The comparison used EPA+DPA+DHA below 8% by weight versus at or above 8%, and triglycerides at or above 100 mg/dL. That triglyceride cutoff is the Fullscript optimal cutoff used here, not a diagnosis of hypertriglyceridemia.

The nearly twofold comparison also needs some perspective. The association was modest, its strength differed between Fullscript and Rupa, and the omega-3 comparison groups were quite different in size: 223 profiles at or above optimal versus 3,149 below optimal, consistent with our previous Insights Pulse issue demonstrating how common sub-optimal omega-3 status is.

Higher omega-3 status also tended to accompany lower hsCRP among 3,202 co-tested profiles. That association was modest but similar across the available source and demographic checks. It does not establish that low omega-3 status causes inflammation. Higher omega-3 status tended to accompany lower hsCRP; this analysis cannot establish that changing omega-3 intake alters inflammation.

Why It Matters

The triglyceride finding fits with the known role of omega-3 fatty acids in lipid metabolism and suggests that omega-3 status may provide useful context when interpreting a lipid panel.

For providers, that could mean looking at diet and current supplement use alongside triglyceride results, particularly when the metabolic picture is not fully explained. An omega-3 result adds context, but it does not replace standard lipid testing.

This analysis did not evaluate supplement doses, treatment responses, or cardiovascular outcomes. It therefore cannot tell us whether changing omega-3 intake would improve triglycerides or who would benefit from supplementation.

Want help explaining this to your patients?

A patient friendly version can be found in our blog entitled, “Your Omega-3 Level Has a Place in the Bigger Metabolic Picture”.

Other Observations: Not Every Expected Connection Was Strong

Some relationships that might be expected based on physiology were weak in these data. Serum magnesium had very little relationship with HbA1c (ρ = 0.045). Ferritin also showed only weak relationships with glycemic markers, while vitamin D had weak associations with hsCRP (ρ = -0.084) and triglycerides (ρ = -0.101).

That does not mean magnesium, iron status, or vitamin D are unimportant in individual care. It means one marker cannot necessarily be used to predict or assume the status of another just because the two are connected biologically. Looking at the actual result still matters.

Why This Research Matters

Providers already use physiology to make sense of laboratory results. Looking at real-world Fullscript and Rupa data gives us an opportunity to see how those relationships appear across tens of thousands of tested profiles, including where the patterns are weaker or more complicated than expected.

The point is not simply that some biomarkers move together. It is that different companion results can change how the same finding is interpreted. Ferritin alongside ALT raises different questions than ferritin alongside TIBC. Folate and B12 help explain homocysteine, but kidney function adds another important piece. Omega-3 status can add nutritional context to a lipid assessment without explaining the whole metabolic picture.

For the next lab review, a useful question is: What other information would help me interpret this result, and would it change the assessment?

That does not mean every unexpected result requires a larger panel. It means using the information already available, and ordering additional testing when it would answer a meaningful clinical question. When broader testing is clinically appropriate, Fullscript Labs can help practitioners use pre-built or custom panels to assess related markers across multiple physiologic systems in one place. The findings in this article do not determine whether additional testing is appropriate for an individual patient. 

How to Interpret These Findings

• These are selected testing groups: Each relationship includes only profiles in which both biomarkers were measured. We do not know why the tests were ordered, so the findings should not be generalized to all patients or treated as population prevalence.
• Association does not mean cause: A prevalence ratio shows how common a finding was in one group compared with another. It does not show that one marker caused the other or predict who would benefit from treatment.
• The cutoffs affect the percentages: A result outside the Fullscript optimal range does not necessarily mean outside the testing laboratory’s reference range, and these findings do not establish diagnostic criteria or treatment targets for an individual patient.
• Clinical context still matters: Diet, supplement use, medications, diagnoses, fasting status, and laboratory differences were not directly accounted for. Age, sex, and data source were examined where information was available, but these checks do not remove those limitations.

Study Notes

• Dataset and Period: De-identified Fullscript and Rupa laboratory records from January 1, 2023 to March 31, 2026 were included. The analysis covered 96,797 patient profiles and 39 pairings among 17 biomarkers. A patient profile is an account-level identifier and may not always represent one unique person.
• Design: This was a retrospective, cross-sectional analysis. For each biomarker pair, we used the earliest testing episode in which both markers were measured on the same date. Different comparisons may therefore include different profiles and testing dates, and the analysis does not track changes following treatment.
• Analysis: Association strengths were assessed using Spearman’s rho. Selected comparisons then used Fullscript optimal ranges to compare how often a related marker was above or below optimal, reported as prevalence ratios with 95% confidence intervals.
• Sources and Demographics: Featured ferritin comparisons used Fullscript records with known sex because the ferritin ranges were sex-specific. Other featured comparisons included Fullscript and Rupa records. Source-specific analyses were also examined when relationships differed between the two datasets. Demographic information was not available for every testing episode. Adult ranges were applied to age-unknown results as a working convention, so some younger profiles may be included.

Learn More

Clinical Insights Pulse is an ongoing series exploring real-world clinical patterns across the Fullscript network. By combining large-scale de-identified clinical data with evidence-informed interpretation, the series aims to help practitioners better understand the patterns emerging in everyday practice and the questions that deserve a closer look.

Labs on Fullscript classify results using clinically informed optimal ranges alongside laboratory reference information, giving practitioners an additional framework for interpreting results within the context of personalized care.

Learn more about Labs on Fullscript and explore how clinically informed laboratory testing can support personalized patient care.

About the Data Behind Clinical Insights Pulse

Clinical Insights Pulse uses de-identified data prepared in accordance with HIPAA requirements. Researchers work only with aggregated datasets, not individual records. All findings describe aggregated clinical patterns, not individual outcomes. Each publication is reviewed before release.

These analyses are conducted in accordance with Fullscript's applicable Terms of Service, Labs Terms of Service, and Privacy Statement, which describe how de-identified information may be used to improve our platform, advance research, and generate clinical insights.

Appendix: Full Biomarker Association Results

The primary analysis assessed association strengths between 39 pre-specified biomarker pairs using Spearman’s rho (ρ). Positive values indicate that the biomarkers tended to increase together, while negative values indicate that one tended to decrease as the other increased. Values closer to zero indicate weaker relationships.

For selected relationships, we also grouped results using Fullscript optimal ranges and compared how often a specified finding occurred between groups using prevalence ratios.

Primary Analysis: Association Strengths

Secondary Analysis: Prevalence Comparisons

For selected relationships, results were grouped using Fullscript optimal ranges. A prevalence ratio (PR) compares how common the specified outcome was in the comparison group with how common it was in the reference group. A PR of 2, for example, means the outcome was twice as prevalent in the comparison group.