Diagnostic Test Accuracy and Clinical Utility Calculator (by www.clinicalutility.co.uk)

	B	C	F	G	H	I	J	L	M	O	P	Q	R	S	U	V	AA	AB	AC
1			allow 5s to calculate.........
2		Diagnositic & Clinical Utility Index Calculator
3		Quickly find the Qualitative and Quantitative Accuracy & Utility of Diagnostic and Screening Tests									try our	emotionthermometer screener
4											share	bit.ly/cuisheet
5		Instructions: enter your data from diagnostic/screening test results in yellow cells and all others are automatically calculated // help? see www.clinicalutility.co.uk
6
7
8	Data Entry							Data Table							Reverse LookUp
9																		0	Very Poor
10																		0.36	Poor
11		Enter Total # Cases:	60						CASES	NON-CASES					If prevalence is	0.15		0.49	Fair
12		total by criterion standard	cases															0.64	Good
13		# Cases Correctly Detected:	59	>>>>		1		TEST POSITIVE	59	4		63	>>>	PPV= 0.937	and sensitivity is	0.886		0.81	Excellent
14		number of true positives	true positives			False negatives			true positives	false positives		total positives
15		Enter Total # Non-Cases:	43					TEST NEGATIVE	1	39		40	>>>	NPV= 0.975	and specificity is	0.985
16		non-cases (aka controls) by criterion standard	non-cases						false negatives	true negatives		total negatives						0	Tiny
17		# Non-Cases Correctly Detected:	39	>>>>		4			60	43		103			and sample is	276	1	13	Very small
18		number of true negatives	true negatives			False positives			total cases	total non-cases		total tested			...............then		2	87	Small
19															Cases	41.4	3	477	Moderate
20	......therefore you had a tested population of		103						Sensitivity= 0.983	Specificity= 0.907		Prevalence =0.583			TP =	36.6804	4	2888	Large
21			0.971												Non-cases	234.6	5	19877	Very large
22															TN =	231.1
23	Quantitative Results							Qualitative Interpretation
24	allow 5s to calculate.........
25					Lower 95%CI	Upper 95%CI
26
27		Sensitivity =>	98.33%		95.09%	100.00%	sensitivity (useful if you already know the number of cases)				The SENSITIVITY of the test is	HIGH…..meaning a positive test often occurs in those with the condition
28		TP / TP+TN
29		Specificity =>	90.70%		82.02%	99.38%	specificity (useful if you already know the number of non-cases)				The SPECIFICITY of the test is	HIGH…..meaning a negative test often occurs in those without the condition
30		TN / TN+FP
31		Positive Predictive Value = >	93.65%		87.63%	99.67%	positive predictive value (useful when you know the number of positive results)				The PPV of the test is	HIGH…..meaning false positives are uncommon in those who screen positive
32		TP / TP+FP
33		Negative Predictive Value = >	97.50%		92.66%	100.00%	negative predictive value (useful when you know the number of negative results)				The NPV of the test is	VERY HIGH…..meaning false negatives are rare in those who screen negative, suggesting a potentially useful screening test
34		TN / TN+FN
35		CUI Positive	0.921		0.861	0.981	case-finding value (ruling-in true positives)				CLINICAL UTILITY of the test for case-finding (confirmation) is	EXCELLENT for case-finding
36		SN*PPV
37		CUI Negative	0.884		0.822	0.947	screening value (ruling out true negatives)				CLINICAL UTILITY of the test for screening (ruling out) is	EXCELLENT for screening (ie confirming a non-case)
38		SP*NNP
39		Prevalence = >	0.583		n/a	n/a					The PREVALENCE of the condition is	BALANCED…..hence close to an ideal prevalence for diagnostic testing
40		CASES/POPULATION
41		Likelihood Ratio +ve =>	10.571		4.15	26.90					The total SAMPLE SIZE is	Small
42		Sensitivity / 1-specificity
43		Likelihood Ratio -ve =>	0.018		0.00	0.129					Notes and cautions	LOW total sample size, be aware of all confidence intervals
44		1-specificity / sensitivity
45		Test Score (Fraction Correct) % =>	95.15		91.00	99.30					Value of this single test for BOTH screening and case-finding is	EXCELLENT (but depends on prevalence)
46		TP+TN/POPULATION
47					Added NPV Value	Added PPV Value					Take home message	This test is EXCELLENT for case-finding (ie confirming a case)
48		Graphical		2.5%	55.8%	35.4%	6.3%				and its best feature is its	sensitivity (useful if you already know the number of cases)
49											We would rate this test as	9/10
50	Normalized Results (per 100 people)							Live Visual (tab #2 to view)
51
52			CASES		NON-CASES
53
54		TEST POSITIVE	57		4		61
55			true positives		false positives		total positives
56		TEST NEGATIVE	1		38		39
57			false negatives		true negatives		total negatives
58			58		42		100
59			total cases		total non-cases		adjusted
60
61
62	Normalized Results (per 1000 people)
63
64			CASES		NON-CASES
65
66		TEST POSITIVE	573		39		612
67			true positives		false positives		total positives
68		TEST NEGATIVE	10		378		388
69			false negatives		true negatives		total negatives
70			583		417		1000
71			total cases		total non-cases		adjusted
72												Open Graphic
73
74	Everday description
75
76
77		In a sample of 1000, where 583 are cases, this test would correctly identify 573 of 583 true cases but it would miss 10 cases (ie false negatives).
78
79		Of the 417 non-cases, this test would correctly identify 378 non-cases but it would falsy identify 39 non-cases as false positives.
80
81		If you ran this test 1000 times, 612 tests would be positive, but of these 39 would be false positives, which is 6.3%
82
83		If you ran this test 1000 times, 388 tests would come back negative, but of these 10 would be false negatives, which is 2.5%
84
85		If you ran this test 1000 times, 951 tests would be correct, and 49 would be false (incorrect) which means the fraction correct would be 95.1%
86
87		but take note of the ...LOW total sample size, be aware of all confidence intervals
88
89		and be aware that even an excellent test may require definitive 2nd stage lab test if the implications of the condition are serious
90
91
92
93
94
95
96
97
98
99
100