HA 640 Unit 3 demand forecast example

Thirty-Six Months, a Winter Peak and a Slow Climb: A Seasonal Demand Forecast for Routine Chemistry Tubes, Tested Against 2025 Before It Was Trusted

Student Name

Health Care Administration, Herzing University

HA 640: Healthcare Operations Assessment and Decision-Making

Instructor Name

January 29, 2026

Thirty-Six Months, a Winter Peak and a Slow Climb: A Seasonal Demand Forecast for Routine Chemistry Tubes, Tested Against 2025 Before It Was Trusted

The Series

The bottleneck assessment found that the specimen processing bench can handle about 130 tubes an hour against a morning peak of about 152. Whether that gap will grow or shrink depends on future volume. This forecast projects routine chemistry tubes for 2026 from 36 months of history, January 2023 through December 2025, taken from the laboratory information system. Volume is expressed as the average number of tubes per day in each month, which removes the effect of months having different lengths.

Data Changes Flagged, Not Smoothed

Two changes affect the series. In July 2024, emergency department specimens moved to a separate priority path and stopped appearing in the routine series. The months before July 2024 were restated by removing specimens with an emergency department location code, so all 36 months measure the same thing; the restated months are marked in the source file. In October 2023, the information system was down for 31 hours over two days; tubes processed during downtime were entered later with estimated times, so October 2023 is averaged over the 29 days with complete records. No other gaps were found. One counting rule is also stated: tubes later rejected for labeling or quality problems are counted as received, because they consume processing time at the bench whether or not they are tested, and such pre-analytic problems are the most common source of laboratory error (Plebani, 2006).

Pattern and Method

Two patterns are visible. Volume is highest from November through March and lowest in July, the pattern expected from higher inpatient census in respiratory season. And the annual average has climbed from 720.4 to 754.1 tubes a day over two years, about 2.3% a year. A method that ignored either would fail: a simple moving average would lag the winter rise and misstate the summer trough, and a straight trend line would miss the season entirely.

The method chosen is classical decomposition with a multiplicative seasonal index and a linear trend. It is transparent, can be checked by hand, and suits a series with a steady trend and a stable seasonal shape. More complex methods are not always more accurate for hospital volumes: a comparison of methods for daily emergency department volumes found that a regression on calendar patterns performed as well as or better than more elaborate models (Jones et al., 2008). The method follows the standard approach set out in a widely used forecasting text (Hyndman & Athanasopoulos, 2021).

Calculation, Shown for January 2026

Step 1, seasonal index. For each year, divide January's value by that year's annual average: 780 ÷ 720.4 = 1.083 for 2023; 778 ÷ 737.1 = 1.056 for 2024; 804 ÷ 754.1 = 1.066 for 2025. The average is 1.068. The twelve monthly indices are then scaled so that they average exactly 1.000; January's index remains 1.068 after rounding.

Step 2, trend. Each month's value is divided by its seasonal index to remove the season, and a least-squares line is fitted to the 36 deseasonalized values, with months numbered 1 to 36. The fitted line is 713.76 + 1.267 × month number, meaning about 1.3 tubes a day added each month.

Step 3, forecast. January 2026 is month 37. Trend value: 713.76 + 1.267 × 37 = 760.6. Forecast: 760.6 × 1.068 = 812 tubes a day.

Testing Accuracy Before Trusting It

The method was fitted to 2023 and 2024 only and used to forecast each month of 2025, which was then compared with what actually happened. The mean absolute percentage error was 2.1%. For comparison, a naive forecast that simply repeats the same month of the previous year had an error of 3.4%. The method beats the naive benchmark by a meaningful margin, Its largest single miss was March 2025, forecast at 781 against an actual 746, an error of 4.7%, when the respiratory season ended early; the core winter months, December through February, were each within 1.6%, which suggests the method handles the peak adequately.

Projection for 2026, With an Error Range

Refitted on all 36 months, the method projects the following average tubes per day. The error range is plus or minus 27 tubes a day, about 1.96 times the 13.9-tube standard error of the fitted model, an approximate 95% range that assumes future errors behave like past ones.

What the Forecast Means for the Constraint

The morning peak rate is projected to stay above the processing bench's capacity of about 130 tubes an hour in every month of 2026, reaching about 160 an hour in February. Even the July trough, about 139 an hour, exceeds it. The problem found in the bottleneck assessment is therefore not a winter anomaly that will pass, and it will worsen slowly as the trend continues. The forecast assumes the morning draw schedule stays as it is, with 39% of the day's tubes arriving between 5 and 7 a.m.; if the draw schedule changed, the peak rate would change even with the same daily total, which is one of the options examined later. The forecast will be updated quarterly, and any month outside the error range will trigger a review of the method.

References

Hyndman, R. J., & Athanasopoulos, G. (2021). Forecasting: Principles and practice (3rd ed.). OTexts. https://otexts.com/fpp3/

Jones, S. S., Thomas, A., Evans, R. S., Welch, S. J., Haug, P. J., & Snow, G. L. (2008). Forecasting daily patient volumes in the emergency department. Academic Emergency Medicine, 15(2), 159-170. https://doi.org/10.1111/j.1553-2712.2007.00032.x

Plebani, M. (2006). Errors in clinical laboratories or errors in laboratory medicine? Clinical Chemistry and Laboratory Medicine, 44(6), 750-759. https://doi.org/10.1515/CCLM.2006.123