Millions of patients are told to fast before blood tests, but new research suggests strict fasting may matter for only a limited set of routine tests.
Study: Fasting Duration and Differences Across Routine Laboratory Tests. Image credit: Andrey_Popov/Shutterstock.com
The impact of fasting duration before blood collection on analytical performance is analyte-specific and not uniformly applicable to all types of analytes, according to a new study published in JAMA Internal Medicine.
Uncertainty about how long patients should fast
Analytical performance of clinical laboratory tests is vital for guiding clinical decision-making. Measurements of many analytes (chemical or biological substances analyzed in laboratories) are sensitive to pre-analytical conditions, such as recent food intake.
To reduce this variability, patients are instructed to fast for 8 to 12 hours before blood collection. Although this practice is routinely applied to a wide range of analytes, it still remains uncertain whether variations in fasting duration influence routine laboratory test results.
To address this gap in clinical practice, the current study investigated whether variations in fasting duration are associated with clinically meaningful differences in test results across a wide panel of 121 routine laboratory analytes.
Millions of test results reveal fasting patterns
A total of 9,755,547 test results from 101,148 patients across 372 test items were analyzed in the study. All test results were obtained from outpatient blood samples collected at a single secondary-care university hospital in the Seoul metropolitan area of Korea. Of the 372 test items, 121 had sufficient observations to be included in the primary between-group and within-person analyses.
Based on fasting duration before blood collection, patients were categorized into three groups: less than 8 hours of fasting, 8 to 12 hours of fasting, and more than 12 hours of fasting. Most test results were obtained after more than 12 hours of fasting, with 90% of results in this group, compared with 8.3% in the less-than-8-hour group and 1.6% in the 8-to-12-hour group.
The study used three analytical approaches: a between-group comparison, a within-person comparison, and a time-trend analysis at 30-minute intervals within the subgroup fasting for less than 8 hours.
The within-person comparison included patients who underwent the same test under two different fasting durations across separate visits.
Only some analytes respond meaningfully to fasting
The comparison between three fasting duration groups identified 27 analytes for which the maximum percentage difference exceeded the optimal analytical performance specifications derived from biological variation (EFLM Biological Variation Database).
However, only 10 of these 27 analytes were reproduced in within-person comparison, while the remaining 17 either were not confirmed within individuals or could not be reliably estimated because of insufficient paired measurements, suggesting that many of the apparent between-group differences reflected differences between patients rather than fasting duration itself.
Among tested analytes, glucose, triglycerides, γ-glutamyl transferase, and lipase showed higher values with less than 8 hours of fasting, whereas total bilirubin and 25-hydroxyvitamin D were higher with more than 12 hours of fasting.
In within-person comparisons between 8 to 12 hours of fasting and more than 12 hours of fasting, none of the tested analytes exceeded the analytical performance specification. A few analytes, however, showed a within-person difference despite no meaningful between-group difference, including chronic glycemic markers glycated albumin and glycated hemoglobin. Both markers showed higher values with less than 8 hours of fasting.
Since these analytes are cumulative markers of glycemic control over weeks to months, acute changes in their levels with a single fasting interval are very unlikely. The authors therefore interpreted these findings as reflecting differences in patients’ underlying glycemic status between visits rather than an acute fasting effect, illustrating that even within-person comparisons cannot eliminate time-varying confounding.
In the time-trend analysis within the subgroup fasting for less than 8 hours, glucose was the only analyte for which the difference between 0.5 and 4 hours after the last meal exceeded its available analytical performance specification. Lactate also showed a substantial 27.9% reduction over this interval, although no EFLM analytical performance specification was available for lactate.
Most routine tests show minimal fasting differences
The study reveals that the impact of fasting duration on laboratory test results is analyte-specific, and that most routinely tested analytes are minimally affected by fasting duration. Specifically, the study identifies only a few analytes, mainly glucose, triglycerides, γ-glutamyl transferase, and lipase, for which the variations in test results due to changes in fasting duration exceeded the optimal analytical performance specifications.
The time-trend analysis shows that blood glucose decreases by about 4.3 milligrams per deciliter for each additional hour of fasting, and the authors concluded that a conventional 8-hour or longer fasting duration is necessary for accurate fasting glucose measurements and diabetes screening.
Similarly, adjusted blood lactate levels were 27.9% lower at 4 hours than at 0.5 hours after the last meal, consistent with a transient post-meal rise in lactate that subsides over the following hours.
Taken together, the study findings suggest that conventional rigid fasting requirements may warrant reconsideration in favor of more flexible, analyte-specific testing protocols for many clinical situations.
For analytes tested in the study, fasting duration was recorded only when at least one test explicitly requiring fasting was concurrently ordered, such as glucose and lipid profile. As a consequence, analytes without specific fasting requirements were included in the analysis only when they were co-ordered with a fasting-required test. These factors may introduce selection bias.
Other limitations include the study's retrospective, single-center design and its restriction to outpatient samples. The researchers could not verify patient-reported fasting durations or control for meal composition, while within-person comparisons could not eliminate time-varying factors such as time of day, intercurrent illness, or medication changes between visits. The study population also skewed toward older adults, with a median age of 61 years, which may limit applicability to younger populations. Importantly, the study did not directly assess how often fasting-related differences caused results to cross clinically meaningful diagnostic or treatment thresholds.
Future multicenter studies are needed for further validating these findings and supporting the implementation of this analyte-specific approach in clinical practice to reduce patient burden while maintaining test accuracy.
Novel blood test could detect pancreatic cancer at its earliest stages