LEARN BUSINESS INTELLIGENCE

Finding Patterns in Missed Medication Doses

To find patterns in missed medication doses, compare the recorded count with the number of doses scheduled for the same group and period. Consistent inclusion rules and dates make the comparison meaningful.

Follow a fictional ward through missed medication dose analysis, from defining the measure to checking weekday results.

  
SET THE QUESTION

Define What Counts as a Missed Dose

To find patterns in missed medication doses, first agree which records belong in the analysis. A chart can compare recorded events by weekday or ward, but its usefulness depends on a consistent definition of the dose opportunity and its outcome.

We’ll follow a fictional pharmacy reporting team looking back over four complete weeks on Ward A. Its question is: how do recorded missed doses vary by the weekday on which they were scheduled? This is a retrospective review, separate from a worklist of patients needing attention now.

The team defines one dose opportunity as one scheduled administration of a medicine. A prescription may create many opportunities over time. Count each eligible opportunity once, using its identifier and the relevant final status at the reporting cutoff.

For this example, the team includes scheduled doses due during the reporting period on Ward A. It excludes orders cancelled before the dose was due and medicines prescribed only when required. Each remaining opportunity contributes once to the eligible scheduled-dose total.

The example uses fictional outcome labels. “Administered” contributes to the eligible total only. “Not given,” “Patient refused” and “Clinically withheld” contribute to both the eligible total and the recorded missed-dose count. Agree the equivalent source statuses with a pharmacy subject specialist and the source-system owner; another audit may use different inclusion rules.

Keep an unavailable outcome distinct from a recorded missed dose or a recorded administration. Where eligibility is known, retain the opportunity in the eligible total and show the unavailable-outcome count separately. State when outcomes were extracted so readers know the point at which the records were assessed.

Count opportunities, not people

One patient may have several scheduled doses, and one prescription may cover several days.

A patient count is a separate measure with its own definition.

  
COMPARE LIKE WITH LIKE

Read Counts Alongside Dose Opportunities

A count shows the volume of recorded missed doses. A rate puts that count alongside the number of eligible scheduled doses—the denominator. Use the same ward, period and inclusion rules for both.

In this example, all eligible opportunities have known outcomes at the reporting cutoff. The figures are invented to explain the analysis; they are not hospital benchmarks.

Fictional Ward A: four complete weeks, grouped by scheduled weekday
WeekdayRecorded missed dosesEligible scheduled dosesRecorded missed-dose rate
Monday126002.0%
Tuesday181,2001.5%
Wednesday96001.5%
Thursday63002.0%
Friday66001.0%
Saturday93003.0%
Sunday123004.0%

Tuesday has the largest count: 18 recorded missed doses out of 1,200 eligible opportunities. Sunday has fewer recorded missed doses, at 12, but they represent 4.0% of its 300 opportunities. The count and the rate answer different questions.

Calculate the recorded missed-dose rate as recorded missed doses ÷ eligible scheduled doses × 100. The numerator must be a subset of the same eligible opportunities used in the denominator.

For the whole example, add the counts before calculating the rate: 72 recorded missed doses out of 3,900 eligible opportunities, or about 1.85%. Averaging the seven daily percentages would give each weekday equal weight despite their different volumes.

If your extract contains only missed-dose records, you can report counts but cannot calculate this rate from that extract alone. Obtain the matching eligible opportunities or label the view as a count analysis.

Keep the comparison descriptive

A higher recorded rate identifies a difference to investigate.

It does not establish avoidability, cause or the quality of a ward’s care.

  
MAKE DATES MEANINGFUL

Keep Weekdays and Reporting Periods Clear

Use the scheduled dose date to answer this example’s question. An outcome entered after midnight may belong to an opportunity scheduled the previous evening. Grouping by the entry date would answer when documentation occurred instead.

Prepare the weekday from the agreed date after applying the reporting time zone. Keep a readable label and a separate sort value, such as Monday with 1 through Sunday with 7. Check the mapping with known dates rather than assuming that every system numbers weekdays the same way.

Use full weekday names or clear labels such as Tue and Thu. Single letters make Tuesday and Thursday, or Saturday and Sunday, difficult to distinguish. Check that the chart keeps them as separate categories and orders the week as intended.

Four complete weeks give this example four occurrences of each weekday. A partial period may contain more Mondays than Sundays, so show the date range and check how many of each day it contains. For a longer review, inspect the individual weeks as well as the combined weekday view.

Attribute each opportunity to the ward defined for the scheduled time, using the available movement history and agreed rules. A patient’s current ward may differ. Keep unresolved ward or date mappings visible as a data-quality category so records do not disappear silently.

Keep missing distinct from zero

Show a weekday with no eligible opportunities explicitly.

A missing extract or an unresolved date is a different result and needs its own label.

  
BUILD THE ANALYSIS VIEW

Connect the Pattern to Supporting Records

Start with the reporting available in your prescribing system or an approved spreadsheet. If it can preserve the definitions, calculate the measures and let readers check the supporting rows, it may meet the need.

A shared dashboard becomes useful when several readers need the same prepared analysis and a consistent way to explore it. IntelliFront BI is self-hosted business intelligence software from ChristianSteven Software. It provides a browser-based place to build and share dashboards on infrastructure your organization controls, keeping reporting data within your environment.

For Ward A, its chart and grid visuals can present the prepared weekday comparison and supporting dose records. A Combo Box visual can provide a ward selector; the dashboard filter setup guide covers that configuration.

Prepare both the eligible-opportunity data and the recorded missed-dose classification before building the visuals. The reporting team owns those definitions and their source mapping. Connecting a dataset does not determine which clinical statuses belong in the measure.

Design a reader journey from a weekday result to the records that explain it. Selecting Sunday in the example should show 12 recorded missed-dose opportunities, with the matching denominator of 300 available in the summary or supporting eligible-dose view. Apply the ward and period selection consistently to both.

Keep recorded reason codes available where the reader is authorized to see them. Use readable category names and show an unavailable reason explicitly. For Sunday’s 12 recorded missed doses, readers should be able to see how those records are distributed across the recorded reasons.

Display the reporting period, extraction time and measure definition near the view. Configure authorized access separately from viewing filters, and show only the supporting information the reader needs.

Check both sides of a filter

Select a ward and period, then check the missed-dose count and the eligible-dose total.

A numerator that changes against an unchanged denominator can create a misleading rate.

  
CHECK THE INTERPRETATION

Test the View Before Drawing Conclusions

Begin with a small fictional or approved test sample you can count by hand. Under this example’s rules, an administered dose adds one to the eligible total only; a dose marked “Patient refused” adds one to both totals.

An order cancelled before the dose was due adds to neither, while an eligible opportunity with an unavailable outcome adds to the eligible total and the separate unavailable-outcome count.

Then reconcile the seven weekday groups in this example. They should total 72 recorded missed doses and 3,900 eligible opportunities. Tuesday should show 18 out of 1,200, or 1.5%; Sunday should show 12 out of 300, or 4.0%. A combined Tuesday-and-Sunday selection should show 30 out of 1,500, or 2.0%.

Test a dose scheduled before midnight with its outcome entered the next day. Check a patient who changed wards, a corrected outcome and an opportunity with several history rows. Each case should follow the chosen date, ward and status rules without adding duplicate opportunities.

Make missing information visible. An empty result should distinguish no eligible opportunities from unavailable source data. A zero denominator produces an unavailable rate, not 0%; a known denominator with no recorded missed doses can produce 0%, with any unavailable outcomes still shown separately.

Review the pattern across individual weeks before treating a combined weekday result as persistent. Retain the counts and denominators beside rates, especially for small groups. Check whether the combined result is spread across the weeks or concentrated in one.

Agree the next investigation with the pharmacy or service-improvement team. Record the question raised, the records to review and who will follow it up. Keep those follow-up notes with the measure so the next review can build on what you learned.

Start with one ward, one reporting period and a measure you can reconcile. If colleagues need to review the same figures, see how pharmacy reporting dashboards can bring the figures together.

Contributors:

Christian Ofori-Boateng

Christian Ofori-Boateng

CEO, ChristianSteven Software