The Detail
Most glucose time-series graphs place time on the horizontal x-axis and glucose on the vertical y-axis. That convention is common, not guaranteed. A profile might place clock time across the bottom and percent of observations on the side, while a histogram uses glucose bins on one axis and frequency on the other. Read the axis titles rather than identifying the chart by color or by the shape of a line.
A complete graph should identify units and the time window. In the United States, glucose is commonly displayed in milligrams per deciliter, abbreviated mg/dL. Other reports may use millimoles per liter, mmol/L. The numbers on those scales differ substantially for the same glucose concentration. Preserve the unit with every copied value and use a documented conversion only when a comparison genuinely requires it.
Axes Set the Coordinate System
Tick labels show how position maps to quantity. On a linear y-axis, equal vertical distances should represent equal numeric changes. If ticks read 80, 100, 120, and 140 at even spacing, each step represents 20 mg/dL. If a label is skipped or the axis contains a break, that choice should be visible. CDC guidance for graphical data presentation calls for clear scale divisions, units, and mathematically consistent spacing.
The x-axis also needs inspection. Labels such as 6 a.m., noon, and 6 p.m. may represent true elapsed time, or they may be category labels placed at equal distances. Points can be aggregated by five minutes, hour, day, or another interval. If the chart connects widely separated observations without marking a gap, the line can imply continuity the dataset does not contain.
Scale Changes Visual Steepness
Suppose a line moves from 110 to 150 mg/dL. On a graph spanning 80 to 180, the 40-unit change occupies 40% of the plotting height. On a graph spanning 0 to 300, it occupies about 13%. The underlying values and difference are identical, but the first line looks much steeper. Aspect ratio also matters: a narrow, tall plot exaggerates vertical movement compared with a wide, short plot.
A nonzero baseline is not automatically deceptive for a line chart when it helps show detailed variation, but it must be clearly labeled. Bar lengths encode magnitude from a baseline more directly, so truncation can be especially misleading there. For any chart, compare numeric endpoints and tick spacing before describing a change as large, small, sharp, or flat. Those visual adjectives are partly products of the frame.
Units Must Travel With the Value
A rough glucose conversion commonly used between the two systems is mg/dL divided by 18 to obtain mmol/L, with the reverse operation multiplying by 18. A reporting system may use more precise factors and specific rounding. A converted result should be labeled as converted and should retain the original value. Never compare 126 mg/dL with 7 mmol/L as though 126 were numerically larger in a meaningful way; they are approximately corresponding expressions on different scales.
A1C percentage is another distinct scale. An eAG in mg/dL may be calculated from A1C, but an A1C percentage does not belong on a glucose y-axis unless the chart uses a clearly labeled secondary axis. Dual-axis charts require special care because the software can choose separate ranges that make unrelated lines appear to rise together. Trace each line to its own axis and legend.
A Worked Example
Illustrative data, not patient results.Teaching Graph J plots four invented values: 110 mg/dL at 9 a.m., 130 at 10 a.m., 120 at 11 a.m., and 150 at noon. Version J1 uses a y-axis from 80 to 180 mg/dL with 20-unit ticks. Version J2 uses a y-axis from 0 to 300 mg/dL with 50-unit ticks. Both use the same points, time stamps, and unit. J1 looks more dramatic because its vertical window is tighter.
A precise description works from coordinates: the plotted value increases by 20 mg/dL from 9 to 10, decreases by 10 from 10 to 11, and increases by 30 from 11 to noon. That description survives either scale. It does not classify the values, explain why they moved, or recommend a response. The example’s only claim is that axis limits influence appearance.
| Time | Plotted value | J1 axis | J2 axis |
|---|---|---|---|
| 9 a.m. | 110 mg/dL | 80–180 mg/dL | 0–300 mg/dL |
| 10 a.m. | 130 mg/dL | 80–180 mg/dL | 0–300 mg/dL |
| 11 a.m. | 120 mg/dL | 80–180 mg/dL | 0–300 mg/dL |
| Noon | 150 mg/dL | 80–180 mg/dL | 0–300 mg/dL |
Legends, Bands, and Markers
A shaded band may represent a configured range, a percentile interval, an uncertainty region, or hours with no data. Color alone cannot identify it. Read the legend and any footnote, and check whether the boundaries are labeled on the axis. If the report calls the band a target range, find out whether it is a software default, a clinician-set value, or another documented setting before describing its role.
Markers may indicate meals, activity, medication entries, sensor changes, or annotations. They are often user-entered and can be incomplete. A missing marker does not prove an event did not occur, and a nearby marker does not prove it caused a change in the line. Dashed and solid segments may distinguish predicted, smoothed, or observed values. Preserve those definitions when exporting the graph to text.
Gaps, Compression, and Overplotting
A blank segment can indicate missing data, but some charting tools connect the nearest available points. Check the timestamps or hover labels when available. On a long window, thousands of values may occupy fewer horizontal pixels than there are observations. The renderer can select, average, or overlap points. A line that looks solid does not prove uninterrupted coverage.
Zooming may reveal details, but note whether the statistics recalculate for the zoomed region. A mini-map or scrollbar can show that only part of the window is visible. Screenshots should include the axis labels and selected range; cropping to the line removes the coordinate system needed to read it. For accessible sharing, provide a short text summary of axes, window, units, and visible gaps.
What It Does Not Tell You
A graph cannot diagnose a condition, determine a personal target, explain the cause of a change, or direct treatment. Its line represents recorded or aggregated values under stated rules. Sensor characteristics, missing data, time lag, and display processing can all affect the picture. A visually steep segment is not a clinical category, and crossing a software band is not, by itself, a complete interpretation.
The chart also does not show values outside its time window or vertical limits. Clipped points may appear as arrows or flat edges, and absent arrows do not guarantee nothing was clipped unless the scale contains the full data range. When personal decisions are involved, use current device instructions and discuss the complete record with a qualified healthcare professional.
Read From the Frame to the Line
Use a fixed order: chart title; start and end timestamps; x-axis variable; y-axis variable and unit; minimum, maximum, and tick interval; legend; bands and markers; data-coverage note; then the plotted shape. For comparisons, verify that both graphs use the same unit, window, aggregation, and scale. If they do not, describe the mismatch before describing the lines.
Finally, translate visual claims back into coordinates. Instead of the second chart is flatter, write both charts contain the same four values but use different y-axis limits. Instead of the spike is huge, state the start value, end value, elapsed time, and unit shown. Coordinate-based language is less dramatic and more reproducible, which makes it a better foundation for questions about the data.