About 350 Records Is Not 350 Patients: Reporting Early Numbers

Some weeks into a frontline programme in a hospital, the first progress report went to the board. It showed about 350 tracked records and one rate of 54%. Both numbers were real. Both could have been read the wrong way in a single glance.

A board member sees "350" and thinks 350 patients. Or 350 successes. A rate of 54% looks like "the programme is halfway there". None of that was true, and the report said so on the same page: the records were not a patient count, not a success count, and the rate described the records, not the programme's progress.

We start from proof before claim. Early numbers are where that rule is tested hardest, because everyone wants good news and the numbers are small.

Why numbers bend

In 1979 the social scientist Donald Campbell put the problem in one sentence: "The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it was intended to monitor." Nobody has to lie for this to happen. A count that the board celebrates becomes a count people produce.

The UK's Code of Practice for Statistics, written for official statistics, rests on three ideas any institution can borrow: trustworthiness, quality and value. Quality includes being clear about what a figure measures and how sure you are. The code is meant for government statisticians. The habit it describes fits a ward report just as well.

Six rules for the first report

• Name the unit. Say what one record is. "One observation by a coordinator, about one situation" is a unit. "Interactions" is not.
• Count each source once. If the same moment is recorded twice, by two people or in two places, it is still one moment. Merge before you count.
• Give every rate its denominator. A percentage needs a base that makes sense. Fifty-four per cent of what, in which weeks, from whom? If the base is unstable, show the count instead.
• An empty week is "not available". No data is not the same as zero. A week with no observations means nobody looked, or the records did not arrive. Write that, and leave the line blank on the chart.
• Mark what was restored. When older records are brought back into a system, label them restored. Readers should know which numbers were recorded live and which were rebuilt later.
• Show a sequence, not a cause. Put dated interventions on the same timeline as the results. For example, "The guide was revised in week 5; clearer wayfinding was recorded more often from week 6" is honest. "The guide raised wayfinding clarity" is a claim the data cannot carry.

Targets are not results

An eight-week programme for frontline service might set targets like these: clear wayfinding from the first point of contact in 60% of observed situations by week 4 and 75% by week 8; no visitor left without an answer in 60%, then 70%; a visitor updated during a delay in 50%, then 60%. A weekly sample of at least 20 situations, spread across shifts, keeps the base from shrinking to a handful.

Those are targets. Put them on the chart as a line, and put what was observed beside them. When the two differ, the difference is the finding.

What to write next to the number

Three short lines under any early figure prevent most misreadings.

• What it counts, in one sentence.
• What it does not count.
• What would change our reading of it.

It feels slower. It is faster than repairing trust after a board member quotes "350 patients helped" in a meeting.

How Reflection keeps early numbers honest

Reflection follows these rules by design. Each source is counted once, a rate is shown only with a valid denominator, an empty week shows as not available, restored records are marked, and interventions appear as a dated sequence beside the results. Team figures appear only for groups of five or more. People read the numbers and decide what they mean.

Sources

1. Assessing the Impact of Planned Social Change. Evaluation and Program Planning, 1979.
2. Code of Practice for Statistics. UK Statistics Authority, 2018.

Comments

Your nameYour comment

Comments appear after our team reads them. Please do not include patients' or colleagues' personal details.

Post comment