Methods
Results
How the figures of a report are put together and qualified: the scope of a study, the checks before publishing, the spread between procedures, costs, the quality of the data and the uncertainty. The figures on this page come from the reports of two fictitious demonstration studies.
In short
- A procedure study counts per procedure, a component study per use, and the two are never put on one axis.
- The spread between procedures is the mean ± the sample standard deviation.
- Each figure has a range that follows its data-quality rating; the tornado diagram ranks what drives the result.
- Costs sit beside the impact and are frozen in no report.
Methods of ClinImpact 2.21. Reports name the version they were computed with; the changelog says what changed.
Contents
Procedure studies and component studies#
A project has a scope. A procedure study is the default and the model every other section of the methods describes. Procedures are recorded one by one, the total is the sum over them, and the per-procedure figure is their mean. No report can be published before one is recorded.
A component study is a project with no procedures at all, in which materials, anaesthesia or sterilisation are studied on their own, or one product's footprint. Its functional unit is one use of its template: what is entered on the project is counted once, and every figure is "per use". The database refuses procedures on a component study and refuses to turn a project with procedures into one.
A component study is compared only with other component studies. It is left out of a subject's pooled averages and of a subject's comparison, which stay clinical. Its published report, the Standard report and the exports name the unit. The two kinds are never divided by the same thing, so they are never put on one axis.
A project can carry a study scope statement in the study team's words, and a partial scope mark. A partial scope means the study leaves out what the compared alternatives share, so its results are a difference between them and not a complete footprint. Both are frozen with each published version.
The Standard report prints the statement under its goal. For a partial scope it also says so on the cover, in the summary and in the limitations, as does the methods paragraph of the export.
Figure 1 as a table
| Section | Study A (kg CO₂ eq per procedure) | Study B (kg CO₂ eq per procedure) |
|---|---|---|
| Products | 7.53 | 1.89 |
| Waste | 6.04 | 1.02 |
| Room use | 2.06 | 0.20 |
| Anaesthetic gases | 1.39 | 0.00 |
| Sterilisation | 0.92 | 0.46 |
| Transport | 0.45 | 0.09 |
| Drugs | 0.01 | 0.01 |
| Total | 18.40 | 3.67 |
Checks and sensitivity#
Before publishing, the project page flags one line that is more than half of the total under EF 3.1 Climate change, because a figure that rests on one quantity and one factor is worth checking. The Standard report names such a line in its findings and limitations.
Besides the other characterisations, the anaesthetic gases and the waste method, the Standard report's sensitivity table gives the result with the hospital's electricity on another national grid, taking the lowest and the highest of the grids in the catalogue, and the grids of the neighbouring countries (for Belgium: France, Germany, Luxembourg and the Netherlands). These totals are frozen at publication, under the published indicator.
Three more rows are calculated from the frozen lines:
- Room energy at half and at double its value. The equipment energy recorded per procedure is not varied.
- Washer-disinfectors and sterilisers loaded half full. The energy, water and detergent of a cycle are shared over half as many baskets, so the sterilisation lines count twice. Products added to the sterilisation are not varied.
- Without the aftercare. The aftercare counts in the result, so this row shows the result without the drugs and products dispensed for it.
Only electricity that resolves through the hospital's grid moves (sterilisation, room use, and any material recorded as electricity), never the energy inside a background dataset. The project page's grid selector gives the same figure for any one country.
Spread between procedures#
For each project the comparison gives the spread between the project's own procedures. It is stated in the same way everywhere (on screen, in the comparison and Standard reports, the exports and the paper figure), as the mean ± the sample standard deviation. Where a comparison pools projects, a ± between the project means says so. Each line of the project is apportioned to its procedures:
| Lines | Share of a procedure |
|---|---|
| products, drugs with a product, their transport | its units of the product ÷ units consumed |
| room use | its minutes × the room's kWh/h ÷ the project's kWh |
| waste (disposal and transport to the processor) | its weighed kg (the mean where not weighed or excluded) ÷ the project's kg |
| anaesthesia | its own anaesthesia records |
| sterilisation | its DIN baskets ÷ the project's DIN baskets |
By construction, the procedures add up to the project total. A single procedure gives no standard deviation. A comparison needs at least two published variants, and there is no minimum number of procedures per variant.
Clinical equivalence#
A comparison is only meaningful when its variants do the same thing clinically: the functional unit of ISO 14044 has to hold across the arms. Since 2.20.6 a comparison is therefore published under a clinical equivalence statement in the study team's words, with a reference, of up to 1,000 characters.
The database refuses to publish a comparison without one, on both paths: a chosen selection and a study subject's one-click comparison. The statement is frozen with the version and printed under the introduction of the comparison report, in its PDF and Excel exports, on the public link and in the methods paragraph of the export. A new version starts from the previous one's text.
A single study has no such field: with one variant there is nothing to be equivalent to. Comparisons published before 2.20.6 carry no statement and print nothing.
A published comparison freezes each arm's spread under its published indicator at the moment of publishing, and says when more procedures were recorded after the arm's report. The Standard report computes the spread when it is downloaded, and prints it only while the project still has the version's procedures and mean. Otherwise it says that the spread is not shown, and why.
Figure 2 as a table
| Procedure | Study A (kg CO₂ eq per procedure) | Study B (kg CO₂ eq per procedure) |
|---|---|---|
| Procedure 1 | 18.07 | 3.65 |
| Procedure 2 | 18.41 | 3.59 |
| Procedure 3 | 18.13 | 3.82 |
| Procedure 4 | 18.33 | 3.64 |
| Procedure 5 | 19.36 | 3.83 |
| Procedure 6 | 18.12 | 3.50 |
| Mean | 18.40 | 3.67 |
| Standard deviation | 0.49 | 0.13 |
Costs#
A product can carry several prices, one per label (hospital contract, online, list). The cost per procedure on a project and in the comparison is units consumed × the price under the chosen label, or the earliest price recorded, plus the DIN sterilised (over all procedures) × the hospital's cost per DIN.
A reusable instrument's price is divided by its uses, so it counts per use like its footprint. The project page shows the cost per module (products, drugs, sterilisation) and names what has no price, in which case the cost is a lower bound.
A cost is not an impact and is frozen in no report.
Staff#
Staff are a module too, per recorded procedure. Their cost is:
- the surgeon's and the anaesthetist's fee for the procedure (RIZIV/INAMI; for carpal tunnel release, for example, € 235.44 for the surgeon and € 124.87 for the anaesthetist, NIHDI 2026). Since 2.21.0 these are the project's own: a new project starts from the study subject's fees, and a copy keeps its own when they are changed. A project without fees of its own uses the study subject's;
- plus the number of nurses recorded for that procedure × its room time (the procedure time where no room time is recorded) × the hospital's own hourly nursing cost.
The number of nurses has no default. A missing fee, number of nurses, hourly cost or time makes the staff cost a lower bound, and the page says which.
Cost against impact#
The comparison sets cost and impact side by side for the projects ticked. Each project is a point, cost per procedure against impact per procedure.
The abatement cost against a reference project is (cost − reference cost) ÷ (reference impact − impact), in euros per unit of impact avoided. A project whose impact is not lower has no abatement cost, and one that is both cheaper and lower saves money as well. Where a product has no price the costs are lower bounds, and the figure is marked as approximate.
Data quality#
Each figure can carry two ratings, set by the person who enters it:
| Rating | Levels |
|---|---|
| How the figure was obtained | measured (weighed, metered or counted at the hospital, or the manufacturer's own data for this product or machine model: its weight, its consumption per programme, an EPD), sourced (literature, generic databases, averages from other hospitals or countries), estimated (an expert guess, a proxy or a default) |
| How well the background dataset stands for the real material or treatment | exact, close (the same kind of material or process), proxy (a stand-in for something the database does not hold) |
The ratings apply to:
- the material quantities and datasets of each component, and its disposal dataset;
- the count of each product on a project;
- the anaesthesia records and the DIN figure;
- the hospital's rooms, sterilisation figures and waste route.
Some ratings follow from the data. A weighed bin is measured, a manufacturer's place found by lookup is sourced, and one taken from the country's capital is estimated, as is a drug or agent volume converted at the assumed density of 1.00 g/mL.
A report line takes the weakest rating among the figures it was computed from, and a figure without a rating makes the line unrated. For each section the report states the share of the result resting on each level, and the share of the part computed from a chosen dataset per fit. Lines without a chosen dataset (drugs, sterilisation, room use, transport) have no fit.
The ratings are frozen with each version and change no figure. "Before you publish" flags more than half of the result on estimated or unrated figures and, separately, more than 10 % on unrated ones (a missing rating, as opposed to an estimate). There is no single score, because an average of levels would hide which part of the result is uncertain.
Uncertainty#
Each figure has a range. By default it follows the data-quality rating: measured ± 5 %, sourced ± 10 %, estimated ± 30 % and unrated ± 30 % (defaults an administrator can change). A range set on the figure's product or anaesthesia record takes precedence, for example the spread of sevoflurane use by fresh gas flow.
One driver at a time (a product, an anaesthesia record, sterilisation, room use, waste) is moved to either end of its range with the others unchanged. Because the result is linear in every quantity, the total moves by Σ |line result| × range over the driver's lines.
Ranking the drivers by that effect gives the tornado diagram, and moving every driver to the same end at once gives the lowest and highest result. The GWP time horizon is a separate scenario (IPCC 2021 GWP 20) and not a range. The ranges are frozen with each version.
Figure 4 as a table
| Driver | Result (kg CO₂ eq per procedure) | Moves the total by ± |
|---|---|---|
| Waste | 6.04 | 1.81 |
| Custom procedure pack | 3.87 | 1.16 |
| Room use | 2.06 | 0.62 |
| Anaesthetic gas (sevoflurane) | 1.39 | 0.42 |
| Sterilisation | 0.92 | 0.27 |
| Breathing circuit | 0.71 | 0.21 |
| Tourniquet cuff | 0.56 | 0.17 |
| Sterile barrier wrap | 0.52 | 0.15 |
| 27 other products and drugs | 2.33 | 0.70 |