QUALITY ASSURANCE Reference Intervals Lecture 4. Normal range or Reference interval The term ‘normal range’ is commonly used when referring to the range.

QUALITY ASSURANCEReference Intervals

Lecture 4

Normal range or Reference interval• The term ‘normal range’ is commonly used when referring

to the range of values for a particular blood constituent in healthy ‘normal’ persons

• Because of the uncertainty concerning what is ‘normal’, the term ‘reference interval’ has been suggested as more appropriate

• Definition of “reference interval”• The range of values statistically derived from data obtained from a

reference population. The intervals represents the limits of 95% of that population

• Reference range: • The entire data set of values obtained from the reference

population from whom the reference interval is calculated

Importance of Reference interval• Reference population

• The group of individuals from whom the data to calculate the reference interval was obtained

• The reference interval is used by the clinicians for purposes of comparison to determine if a test result is abnormal

• They depend on reference interval to assist them in making judgments concerning a patient’s health

• Therefore, it is the responsibility of the Lab. to provide a reasonably accurate reference interval based on the disease state, test methodology and the patient population served

• Each laboratory should define its own reference intervals for: 1. Instrumentation it uses

2. Patient population it serves

• This is because subtle differences exist between methods and groups of patients

• Therefore reference intervals published by the manufacturers of reagents and instruments should not be used without confirmation

Factor Affecting the Determination of Reference Intervals

1. Composition of the reference population

2. Sample collection and processing procedures

3. Method of analysis

4. Statistical treatment of the data after it has been gathered

Methods of Reference Interval Calculation

• Samples should not be treated as (special), but should be collected and processed the same as the routine laboratory workload

• Unacceptable samples such as hemolysed, contaminated or improperly collected should not be used

• The method of analysis should be carefully monitored including precision, accuracy, quality control and they all should be consistent with the conditions of performing the normal testing

• When method of analysis is changed or altered, new reference intervals should be determined, and this new interval should be presented to the medical staff

• The number of individuals required for reference interval will depend on the method of calculating the reference interval

Methods of Reference Interval Calculation

• The distribution of values within the group will play an important role in: 1. Determining the required sample population size

2. Which method is used to calculate the reference interval

Methods of Reference Interval Calculation

• There are two methods for calculating the Reference Interval which depend on the distribution of the population1. Calculation of reference intervals using Percentile Ranking

• Makes no assumption about the sample populations’ distribution and can be applied to both Gaussian and non-Gaussian distributed sample groups (non-parametric methods)

2. Determining Reference Intervals on Gaussian-distributed Data• Parametric methods

Calculation of Reference Intervals Using Percentile Ranking

• This method involves: 1. Rearranging values in ascending order and selecting

the middle 95% as the reference interval

2. Find 2.5 percentile rank, a) multiply the value n + 1 (n equals the number of values in the

data set) by 2.5% [(n+1) X 2.5 % = ??]

b) Use this number and count from the 1st percentile (lowest value) by that amount. This will be the lower limit

3. Do the same for the upper limit. Count down from the highest percentile to the 97.5 percentile

Example

• No. of values = 100

• 2.5% → (100 + 1) x 2.5% = 2.5 ~ 3

• Third value from the top is 10 s• Third value from the bottom is 12.7 s• The values between these numbers

are the reference range which are 10.0 to 12.7 s in this case

Problem of Outliers• If we narrow the range to 90%, large number of false positive tests may occur

• With this method, outliers can affect the calculation of reference interval

• Outlier is an observation point that is distant from other observations• They can be due to measurement errors• Errors in data recording or data transcription

Problem of Outliers• The problem of outliers increase with small population and

decrease with large sample population• The problem of outliers may be reduced by using the

following

• If the difference between the highest (or lowest) value and the next in the order is more than one third of the reference range, discard the value and then recalculate the percentile and the range

• X1: the first value in the range, X2: the second value• Xn – X1 is the difference between the highest and lowest in the reference range

X2 – X1 >1/3Xn – X1

Problem of Outliers• Example:• Is a value of 10 in a group of 120 values with a reference

range of 10 to 40 an outlier if the next value is 22?

• Therefore, 10 should be discarded

X2 – X1 =

22 – 10=

12= 0.4 > 1/3

Xn – X1 40 – 10 30

Determining Reference Intervals on Gaussian-distributed Data

• First, calculate the mean, median and standard deviation of the data set

• If the mean, median and mode are essentially the same number, then the distribution is Gaussian

• If distribution is Gaussian, define a range of values ± 3SD from the mean and delete any values beyond these limits (to eliminate outliers)

• Recalculate the mean and SD, where the reference interval will be defined by ± 2SD

• Mean = 11.307• Median = 11.3• Mode = 11.3• SD = 0.78• 3SD range = 9.0 to 13.6• 2SD range = 9.7 to 12.9