Report to Scandinavian Society of Clinical Chemistry (NFKK) from The Nordic Reference Interval Project (NORIP)

Establishment of project. *

Project plan *

Data *

Data base *

Data treatment *

Enzymes *

Non-enzymes *

Target CAL *

Data exclusion *

Calculation of reference intervals *

Method *

Partitioning *

Documentation *

Establishment of project.

The project was established in April 1997 by an initiative from Heidi Steensland an Petter Urdal, Ullevål Hospital, Oslo with the aim to establish common reference intervals for the most common used components in serum and plasma for diagnostic purposes in clinical chemistry.

The project group was appointed by the national clinical chemistry societies in the Nordic countries: Peter Felding, Denmark, Leifur Franzson, Veli Kairisto, Finland, Iceland, Per Hyltoft Pedersen, Denmark, Pål Rustad, Norway, Gunnar Skude, Sweden.

On the first meeting, the group elected Pål Rustad as leader and Peter Felding as secretary.

The group agreed to follow a project plan outlined by Peter Felding and Per Hyltoft Pedersen.

Project plan

The plan, as described in the project description with later refinements was:

We would invite Nordic laboratories in general to participate in the project by:

1. Pay about 5000 DKR to the project group.
2. Collect at least 25 reference person serum-, plasma- and full blood samples (inclusion criteria are defined in the project description - see above) and freeze at –80C. The reference persons should be evenly distributed on gender and age. Some of the samples should be collected for analysis at the laboratory and some for submission to a central data bank (7 serum, 2 plasma, 1 EDTA buffy coat, each 1 mL). Each reference person should fill out a questionnaire to collect relevant data for evaluation of reference intervals.
3. Receive 5 control materials on dry ice: CAL, HIGH, LOW, P, X. CAL (serum pool) should be the calibrator of the project, HIGH (concentrated serum pool by freeze drying) and LOW (HIGH diluted ½ with a Na/Ca in water) to evaluate linearity of methods, P is a serum pool from women using oestrogen and X is a serum pool produced for future use.
4. Analyse controls (10 CAL and three of each of the other in at least one series and 10 X in the rest of series if necessary) and samples in one series at the laboratory. The labs were also encouraged to analyse fresh serum and plasma with controls in series as described above.
5. Submit analytical data, method data, reference person data to a central data base.

Data

Data base

The data are stored in a MS Access relational data base at Fürst Medical Laboratory, Oslo and is administered by Pål Rustad.

Data treatment

The enzymes and the non-enzymes are treated differently:

Enzymes

Heidi Steensland have had the responsibility to select the methods with necessary quality (compatible to IFCC methods at 37C). If the laboratory have used a slope/intercept correction to the submitted data, the data have been transformed back to original values. These data have been used for calculation.

Non-enzymes

The data submitted are all multiplied with the factor Target CAL / Mean CAL measured by the laboratory in that series (or Target X / Mean X if only X have been used in series).

Target CAL

The target for CAL is established in three different ways depending on component (see table here):

1. Transferred value from IMEP 17-1 to CAL by The Nordic Trueness Project, 2002 (IMEP 17 study)
2. Reference method values established in 1997 by DGKC (page 1, page 2, page 3)
3. Median from all laboratories in NORIP

Data exclusion

Data have been excluded for different reasons (see table) :

1. For labs with more than one series: Data are lacking on which controls have been measured with which samples.
2. Same samples measured by different methods on same component.
3. Material difference: Automatic test of extreme differences between combinations of serum, plasma, fresh, thawed - the extreme value is excluded.
4. Person exclusions: Extreme values for one or more components for one person excludes all results for that person, see Peter Feldings Iceland presentation. (glucose > 11 mmol/L, glucose > 7 mmol/L and fasting > 8 hours, 5s/3s and 4s/4s rule: At least two values for different components for one person outside the limits excludes all person results (s is total biological variation based on reference intervals from Malmø/Odense, logaritmic transformations).
5. Method exclusions (not compatible enzyme IFCC 37C-methods, plasma with bad correlation with serum on method).
6. Component specific exclusions: Fasting (triglycerides, glucose), diabetes in family (glucose), physical activity (CK), oestrogen use (TIBC)
7. Enzymes: Results outside mean + 4s (s - standard deviation of gender partitioned distributions using log-transformations)
8. Refval 4.0: Automatic exclusions for non-enzymes (see "Calculation of reference intervals").

Method

Non parametric method based on 2.5- and 97.5 percentiles of distribution with computer program Refval 4.0 based on IFCC recommandations.

Partitioning

Based on theory outlined by Ari Lahti (Clin. Chem. (2002), 48:2, 338-352, presentation Iceland 10/8-2002) and incorporated by him in Refval 4.0, the criteria for not partitioning is <4.1% and >0.9% of each of the subdistributions should be outside 2.5- and 97.5 percentiles of the common distribution.

Gender partitioning is straight forward using this program. Age limits have been estimated by qualified guessing and partitioning program have evaluated if differences are important.

Documentation

Results from the project have been continuously updated on the project web home site http://www.furst.no/norip.

Specific details of each component can be viewed by selecting the specific component from the table on the web page "Compiled data for each component" and a summary for each component is presented here

The proposed reference intervals:

Proposed reference intervals for enzymes

Proposed reference intervals for non-enzymes

Evaluation of reference intervals

Petter Urdal, Ullevål, Norway have taken the initiative to let 7 groups from different laboratories in Norway get the responsibility to evaluate the reference intervals for all components proposed by the NORIP project as presented on the home site. The results of the evaluation will be presented at a one day meeting on 7. April 2003. It is reasonable to take into account the conclusions of these groups before the NORIP group make their final recommendations.

Pål Rustad

Project leader

15/2-2003