1+1 =

Dr Bill Bartlett
Joint Clinical Director
Diagnostics Group
Ninewells Hospital & Medical School
NHS Tayside
 Basics
 Memorise a few formulae
 Read the question
 Check that your answer is credible
(estimate. e.g. half life)
Question 11
In healthy subjects the average within subject biological
variation (CVI) for creatinine in serum is 4.1% and the
between subject (CVG) 12.9%. Assume when answering
the following that the analytical imprecision of creatinine
assays at a value of 70 µmol/L across the UK is 12.2%
and that there is no assay bias between laboratories.

a. What is the analytical goal for imprecision based on biological

variation?

b. Calculate the index of individuality for creatinine.

c. Does the index of individuality support the application of

population based reference intervals?

d. If the total error included only 1 analytical standard deviation

what would be the expected range of creatinine
concentrations seen across the UK in subjects having a mean
creatinine concentration of 70 umol/L.
 Components of Variance in
Clinical Chemistry
Measurements

 Analytical variance. CVA

 Within Subject biological variance. CVI

 Between Subject biological variance. CVG

 Setting of analytical goals.
 Accepted analytical goal for imprecision: -
CVGoal = ½ CVI

therefore: -
CVAnalytical = CVGoal

= ¼ of the total Individual if achieved.

(Harris. Am J Clin Pathol 1979:72;274)

a. The analytical goal for imprecision based on biological variation?

CVI/2 = 4.1/2 = 2.05%
Biological Variation &Utility of Reference
Intervals
Index of Individuality and Assessing
the utility of reference intervals.
 Utility of population based reference data?
 Ratio of Within to Between subject variances.
Index of Individuality = CVI / CVG
b. index of individuality = (CVI)/ (CVG) = 4.1/12.2 = 0.31

 Population Ref Intervals: -

Index <0.6 = Limited in Value
Index >1.4 = Applicable
c. index of individuality < 0.6 so population based ref
intervals of little utility.
 Biological Variation

Gowans & Fraser. Ann Clin Biochem 1988:25:259-263

If the total error included only 1 analytical
standard deviation what would be the expected
range of creatinine concentrations seen across
the UK in subjects having a mean creatinine
concentration of 70 umol/L.

Total error = CVI + CVG + CVA = 4.1 + 12.9 + 12.2 = 29.2%

D) Therefore range around 70 µmol/L = 50 to 90 µmol/L

Question 8
Specimens were taken from a patient for measurement of
a drug that has a half life of 56 hours. Three specimens
were taken for assay on cessation of the drug and received
in the lab labelled time 0, 74.1, and 97.3 hours. The assay
produced gave results of 65, 20, and 15 mg/L respectively
for the sequential samples. The requesting physician
contacts the lab indicating that the patient was a subject
in clinical trial and that the elimination profile looked
odd. Is he correct in his assessment? The critical point
for his data set is the result at 74 hours. The samples are
not available for re-assay as they were accidentally
thrown away post assay and he is not very happy. Is the
situation salvageable?
 Essence of Question
The question simply asks the question as to
whether the results are likely to be correct,
and if not where does the problem lie, and can
we retrieve anything form the situation.
 Information Given
 The usual half life (t) is 56 hours
 Measured concentrations of the drug: -
– 0 Hours = 65 mg/L (C0)
– 74.1 hours= 20 mg/L (C74.1)
– 97.3 hours = 15 mg/L (C97.3)
 First step?
Do the results fit mathematically.
We know that Ct = C0 . e-kt ……… (1)
k= elimination constant
t= time
Ct = Concentration at time t
C0 = Concentration at time 0
What is the value for k?
 Value of k
elimination (k) constant for the drug can be determined as
we know that at one half life t = 56 and the ratio of Ct to C0
will = 0.5: -

Ct/C0 = e-kt …………………….. (2)

or
ln (1/( Ct/C0 ))= kt ……………………..(3)

Therefore; -
ln 1/( Ct/C0 )= ln 1/(0.5) = ln2.0 = k x 56
therefore
0.693/56 = k = 0.01237
Knowing k and t the measured
concentrations can be compared with the
theoretical: -
If it is assumed that the measured concentration at
time 0 is correct then C0 = 65 and t = 74.1
Substituting into equation 1: -
Ct = 65 . e -0.01237 x 74.1
Ct = 26 mg/L
Theoretical concentration at 74.1 hours assuming
measured concentration at time 0 is correct =
26mg/L
Similarly assuming measured concentration at time
0 is correct and t = 97.3
Substituting into equation 1: -
Ct = 65 . e-0.01237 x 97.3
Ct = 19.5 mg/L
Theoretical concentration at 74.1 hours assuming
measured concentration at time 0 is correct =
19.5 mg/L
 Measured v Theoretical
Time (Hours) Measured Calculated
(mg/L) (mg/L)
0 65

74.1 20 26

97.3 15 19.5

Values are discrepant could the baseline be the incorrect measurement?

Back calculate from measured values?
 Back calculate from time t
Assume the measured concentration at t= 74.1 is correct and k =
0.01237
Substitute into equation (1): -
20 = C0 . e -0.01237 x 74.1
C0 = 50.0 mg/L
Assume the measured concentration at t= 97.3 is correct and k =
0.01237
Substitute into equation (1): -
15 = C0 . e -0.01237 x 97.3
C0 = 49.9 mg/L
Baseline measurement incorrect? Balance of
probabilities 74 minute measurement OK
 Question 6
A reference interval has been constructed for a new
analyte. 1000 patient samples were analysed and the
values were found to be normally distributed
following application of appropriate statistical tests.
Calculate the theoretical number of patients that
may be expected to have values that fall within the
following ranges within the population tested: -

– -2 and +2 standard deviations

– over +3 standard deviations
– below -1 standard deviations
– within the coefficient of variation
Basic Statistical Knowledge

Mean ± 1SD = 68 % of the population

Mean ± 2SD = 95.4 % of the population
Mean ± 3SD = 99.6% of the population
 Simple arithmetic
Therefore.

a) number of subjects % with results between ± 2SD =

1000/100 *95.4 = 954

b) %population outside 3SDs =100 - 99.6% = 0.4%

Therefore percentage over +3 SDs = 0.4/2 = 0.2%
Number of subjects = (1000/100) x 0.2 = 2
c) % of population below 1 SD = (100-68)/2=
16%
Number of Subjects = 16% of 1000 = 160

d) The CV is the SD x 100/mean. As we

know ±1 SD contains 68 % of the 1000
subjects = 680 would be within the CV
 Question 14
Calculate the freezing point of a 0.9%
solution of sodium chloride. The cryoscopic
constant water is -1.86 C (1.853 K-
Kg/mol).
 Cryoscopic Constant

In thermodynamics, the cryoscopic constant,

Kf, allows one to relate molality to freezing
point depression. It is the ratio of the latter to
the former: -
ΔTf = Kf · m · i

where i is the van 't Hoff factor, the number of

particles the solute splits into or forms when
dissolved. (e.g. NaCl i = 2 as it splits to Na+ and
Cl-) (serum
A 1 molal solution boils at 0.52°C higher
and freezes 1.86°C lower than pure water.
 Basis of freezing point
depression osmometry.

Therefore a 1 molal solution of osmotically

active particles will depress the freezing point
of water from 0C to -1.86C.
 NaCl Ionises
In solution NaCl ionises according to the
following formula to give 2 osmotically active
components: -
NaCl Na+ + Cl-
(0.155M) (0.155M) (0.155M)
0.310 Moles of osmotically active components.
Step 1: Molar Concentration NaCl
In this problem the task is to determine the freezing
point of a 0.9% NaCl solution. The molar
concentration of NaCl is calculated as follows=
0.9% NaCl = 0.9 g/100 mL = 9 g/L
MW of NaCl =58.5
[NaCl] = 9/58.5 = 0.155 M
Molarity not Molality therefore an approximation
 Freezing point
Freezing point t
= 0C + (0.310 Moles x Cryostatic Constant)
= 0C + ( 0.310* -1.86) = -0.58C
 Check calculation?

Check Calculation

mosmol/L H20 = -0.58/-1.86 * 1000 = 311.82

 Question 12
In a population of 20,000 men to be screened for prostate
disease, using single PSA measurements, there is a
known prevalence of prostate cancer of 3%. The
diagnostic sensitivity and specificity of a PSA value
exceeding 4µg/L for the presence of the cancer is 67%
and 97% respectively. Derive some figures from these
data that may help persuade your local urologist that
using PSA measurements in this way may adversely
effect the size and usefulness of his outpatient clinics.
 Bread & Butter Question
Formulae for the following should be second
Nature; -
 Diagnostic sensitivity
 Diagnostic specificty
 Diagnostic efficiency
 Predictive Values
 Sensitivity of 67% means that 33% of
diseased population will be false negatives
 Specificity of 97% means that 3% of the
non-diseased population will be false
negatives.
 3% of the population of 20,000 tested will
have the disease i.e potentially 600 men.
 Contingency Table based on
20,000 subjects

PSA > 4,ugIL PSA < 4,ugIL Totals

(Positive) (negative)

With Disease 402 (TP) 198 (FN) 600

Sensitivity
67%
No Disease 582 (FP) 18,818 (TN) 19,400
Specificity
97%
Totals 984 19,016 20,000
PPV = TP/(TP+FP) = 402/(402 =582) = 41%
NPV = TN/(TN+ FN) = 18818/(18818+198) = 98.9%
Missed cases = 33% Sensitivity only 67%

 Bottom line for clinics is that 59% of patients seen

would not be diseased
 33% of patients would be missed.
 For consideration
 Impact would depend on annual incidence.
 Existing cases at start up
 Take up of screening
 Comparability of study population with the
population served
 Question 15
The reference data for an analyte was found to
be normally distributed and a reference interval
(mean 2 standard deviations) was determined
to be 120 - 160 mmol/L for a study population.
What would be the impact upon the reference
interval for this population if the assay
imprecision was reduced from a coefficient of
variation of 5% to 2%.
 Understanding Components of
Variance
Range of 120 to 160 = mean +/- 2SD therefore 160-
120 = 4SD
40/4= 10 = 1SD
The total CV of the measurement will comprise of
the analytical imprecision ( CVAnalytical) and the
biological variation (CVBiological) inherent in the
population: -
CVTotal = CVAnalytical + CVBiological …… (1)
 Estimate the Inherent BV
From the information given the CVTotal can be
calculated at the mean value of 140 mmol/L: -
CVTotal = (1 SD / Mean) x 100 = (10/140) x 100 = 7.14%
You are told that the CVAnalytical = 5% therefore
rearranging and substituting equation (1)
CVBiological can be estimated: -
CVBiological = CVTotal - CVAnalytical = 7.14 - 5 = 2.14%
 New CV
If the analytical imprecision is reduced to
CVAnalytical is reduced to 2% then substituting
into (1): -
CVTotal = CVAnalytical + CVBiological
CVTotal = 2 + 2.14 = 4.14
 Predicted reference range
The impact on the reference limits would therefore be as
follows: -
CVTotal = 4.14% so 1SD = (mean/100) x CVTotal
=(140/100) x 4.14 = 5.8 mmol/L

Mean ± 2 SD = 140 ± (2 x 5.8) = 140 ± 11.6

The reference interval (Mean ± 2 SD) = 128.4 - 151.6

mmol/L if the assay imprecision is reduced to 2%.
Assay fit for purpose?
 Question 3
A screening test for a lethal, but easily curable, disease with a
population prevalence of 0.002 has a diagnostic sensitivity of
95% and specificity of 99%. Calculate the negative and
positive predictive values and diagnostic efficiency of the
test for a population of 1 million. The cost of this test is
£1.00 per patient and the cost of the confirmatory test is
£10.00 per patient. What would be the impact on total cost of
provision of screening and confirmatory testing if a new
screening test is adopted which has a sensitivity of 99% and
specificity of 97%, but costs £2.00 per test.
 Current State, Prevalence 0.002
95% sensitivity 99% Specificity
Test +ve Test -ve Totals
Diseased 1,900 TP 100 FN 2,000
No Disease 9,980 FP 98,8020 TN 998,000
Totals 11,880 988,120 1,000,000

Negative Predictive Value = TN/(TN+FN) = 988020/(988020+100) = 0.9988

Positive Predictive Value = TP/(TP+FP) = 1900/(1900+9980) = 0.1599

Efficiency = (TP+TN)/(TP+TN+FP+FN) = ((1900+988020)/1000000) x 100 = 98.99%

 Cost of Current State
Cost of testing = cost of screening test Plus
cost of confirmatory tests on screen positives.

Cost of testing = (1,000,000 @ £1.00) +

(11,880 @ £10:00) = £1,118,800.00
 New System, Prevalence 0.002
99% Sensitivity 97% Specificity
Test +ve Test -ve Totals
Diseased 1,980 TP 20 FN 2,000
No Disease 29,940 FP 968,060 TN 998,000
Totals 31,920 968,080 1,000,000

Cost of testing = cost of screening test Plus cost of confirmatory tests on

screen positives.
Cost of testing = (1000000 @ £2.00) + (319,200 @ £10:00) = £2,319,200

Cost per true positive identified = £588.84 in scenario 1

Cost per true positive identified = £1171.31 in scenario 2
20,040 additional false positives with 80 more cases identified
 Question 7
You are given below copies of HPLC
chromatograms of a catecholamine standard solution
and extract of a urine from a patient with
phaeochromocytoma. The standard concentrations
and retention times are given in the Table. The
method uses an internal standard (DHBA) to correct
for losses incurred during the extraction. If the
patient had a urinary output of 2400 mL in 24 hours,
calculate the 24 hour outputs of noradrenaline,
adrenaline and dopamine in this case.
Identify Peaks & Measure
Heights to Determine PHR
Standard RT Concentration
mins (nmol/L)

Noradrenaline 1.9 500

Adrenaline 3.2 150

DHBA 3.9
(INTERNAL STANDARD)

Dopamine 7.1 1500

Measure Peak Heights in Chromatogram A and B

Peak Height Ratio (PHR) = Peak height of the catecholamine peak in mm

Peak height of the DHBA (Internal Standard)

Because the urinary extract and the standard have the same concentration of DHBA
added as internal standard it is possible to quantify the concentration of the
endogenous catecholamine in the urine using the following formula: -

Concentration = Peak Height Ratio of the extract x Standard Concentration

Peak Height Ratio Standard
Concentration Noradrenaline = 1.607 x 500
0.586
Urinary output = concentration x urine volume in Litres
e.g. Urinary output Noradrenaline = 1371 * 2.4 = 3290 nmol/24h

Catecholamine Retention Peak Peak Concentration Urinary

Time (Min) Height Height nmol/L Output
(mm) Ratio nmol/24h

STANDARD
Noradrenaline 1.9 8.5 0.586 500
Adrenaline 3.2 3.5 0.241 150
DHBA 3.9 14.5
Dopamine 7.1 9.5 0.655 1500
EXTRACT
Noradrenaline 1.95 22.5 1.607 1371 3290
Adrenaline 3.15 5.0 0.357 222.3 533.5
DGBA 4.0 14.0
Dopamine 7.1 27.0 1.862 4264 10,233
 Question 10
What would be the calculated pH of an
aqueous solution of sulphuric acid at a
concentration of 2.94 g/L? Assume the acid is
fully ionised. (Atomic weights, S = 32, O=
16, H=1).
 H2SO4 has a molecular weight of 98

Molar concentration = 2.94/98 = 0.03 = 30 mmol/L.

H2SO4 2H+ + SO4-

Therefore [H +] = 2 * 30 mmol/L = 60 mmol/L

pH = -log of the [H+] = log of 6.0 x 10-2 molar

= 1.22
Antilog of -1.22 = 0.06 molar
 Question 1
Calculate the number of mmols of sodium
reabsorbed in 24 hours by the kidneys of a
healthy 70 Kg man. Assume a daily urinary
sodium excretion of 100 mmols. Clearly state
the assumptions you make.
 Requires some General
knowledge and a little arithmetic
 Assume a GFR of 120 mL/minute for an average adult with normal
renal function
 number of litres of plasma filtered per 24 hours (1440 minutes) can
be approximated as follows: -

Volume Filtered = (mL/minute) x 1440 = 120 x 1440 = 172.8 L

1000
 average sodium concentration of plasma is 140 mmol/L,
 approximate amount filtered per 24h is given by: -
140 x 172.8 = 24,192 mmol/24 hours
 daily sodium excretion is 100 mmol therefore the amount
reabsorbed = 24,192 - 100 = 24,092 mmol/L
 Question 5
Injection loops for high performance liquid
chromatography systems are made out of coiled
lengths of stainless steel tubing of known internal
diameter. When filled the loops enable the
injection of a fixed known volume of sample
reproducibly into the system. If an injection loop
is constructed from tubing with an internal
diameter 0.25 mm, and is 0.306 M long, how
many µL of sample would be injected onto the
system?
General Knowledge around units
of measurement
Volume of a cylinder (V) = cross sectional
area (A) x length (L) = π r2 L
Length = L = 0.306M

Diameter =
2r = 0.25 mm
 Volume of tube
 1mL = 1 cm3 or 1000 mm3. Which means that 1
µL must = 1 mm3.
 we know the length of the tubing is 0.306 m = 30.6
cm = 306 mm
 The Internal volume is thus easily established as
indicated above: -
V=A x L = π r2 L
Therefore: -
= (3.147 x (0.25/2)2) x 306
V= 0.049 x 306 = 14.99 mm3 = 14.99 µL
 Question 13
In a spectrophotometer, a coloured solution
gives a transmittance of 77% in cell with a
2cm light path. Determine the percentage
transmission, and absorbance, that would be
obtained if the sample concentration is
doubled.
Absorbance = 2 – log Transmisssion
or A=2-logT
And
Absorbance = eCL
Where
C = concentration , L = length of light path,
and e is a constant
 Process
 Calculate e for existing conditions
 Calculate A for new conditions
 Calculate T for new conditions
 Calculate e
When T = 77%
then
A = 2 – logT = 2 – log77 = 2 – 1.866 = 0.113
If A= eCL
then
e = A/(CL) = 0.113/(1 x 2) = 0.0565
 CalculateA and T for new
conditions
therefore if the sample concentration (C) is
doubled: -
A = eCL = 0.0565 x 2 x 2 = 0.226
As A=2 – logT
logT = 2 – A= 2- 0.226 = 1.774
T = anti-log 1.774 = 59.4 %
If sample concentration is doubled
A=0.226 and T = 59.4%
 Question 9
A diabetic with a 24 hour creatinine clearance
of 25 mL/minute, and a serum creatinine of
200 µmol/L, was found to have an albumin
excretion rate of 25µg/minute. If the 24 hour
urine volume was 2000 mL calculate the
albumin/creatinine ratio in mg/mmol of
creatinine.
 Creatinine clearance = 25 mL/minute
 Serum creatinine = 200 µmol/L
 Albumin excretion =25 µg/minute
 Urine volume =2000 mL

Need to calculate the concentration of

cretinine in the urine
 Creatinine concentration
To calculate concentration of creatinine in urine in
mmol/L:-
CC= (U·V)/P
Creatinine clearance = Urine creatinine (mmol/L) x Rate urine produced (mL/min)
Plasma creatinine (mmol/L)

25 = (U x (2000/1440))/0.2 (NB! Units)

(25 x 0.2)/ (2000/1440) = U = 3.6 mmol/L
Urine creatinine concentration = 3.6 mmol/L
Amount of albumin in 2000 mL =
Albumin excretion x Number of minutes/day
= 25 x 1440 = 36000 µg = 36 mg
Concentration of albumin/L =36/2 = 18 mg/L

Albumin Creatinine ratio = 18/3.6 =

5.0 mg/mmol creatinine