Table 1
Table 2
Table 3
Table 4
Table 5

Table 1 Performance of microbiology near patient tests (NPTs)
 

Author, Year, Country, Type.	NPT Package(s), Operator	Setting, Subjects	Performance Comparitor (quality score)	Performance Results	Provisional Conclusions and areas for further research
Andersen, 1992, Denmark, Streptococcal throat test	Phadirect Strep A. General Practitioners	Primary care: 2 GPs in one practice. n=105 consecutive patients presenting with a sore throat as the primary reason for presentation.	Laboratory Aerobic Culture of charcoal impregnated cotton-tipped wooden swab on 5% de-fibrinated horse blood-agar plates and chocolate agar plates for 18-24hr at 37°c (6)*	Sensitivity 68% (95% CI 48-84%) Specificity 97% (95% CI 91-100%) PPV 90% (95% CI 70-99%) NPV 89% (95% CI 81-95%)	High specificity achieved in primary care setting with non technologist operator. However the results from this study reflect the use of a single aerobic culture as a gold standard (See Wegner et al below)
True, 1986, USA, Streptococcal throat test	Culturette Brand 10 minute Group A ID test Family physicians on research team.	Primary care (US university family practice office). n= approx 280 individuals aged over 3 presenting to family practice with symptoms suggestive of pharyngitis for whom throat culture was ordered.	5% sheep blood-agar cultures incubated in 5% CO2 +atmosphere for 18-24 hours at 35°c (5)	Sensitivity 82% Specificity 82% PPV 91% NPV 92%	Apparently sensitive and specific NPT but these results were achieved in a university research setting and so may not be generalisable.
Wegner, 1992, USA, Streptococcal throat test	a) Culturette Brand 10 minute Group A ID test. b) Testpack Strep A. c) Reveal. d) Ventrescreen Strep A. e) Cards test. Primary Care Nurse	Primary care - US community office. Subjects all had symptoms of streptococcal pharyngitis. Each of 5 practices had one of the NPTs. The number of patients in each case was: a) Culturette n=176 b) Testpack n=186 c) Reveal n=127 d) Ventrescreen n=143 e) Cards n=123	Double Culture i) Trimethoprim-sulfamethoxazole blood-agar plate incubated anaerobically ii) 5% sheep blood trypticase soy agar plates incubated aerobically. Colony count at 1&2 days. Typed with Group A identification latex reagent. A culture with any quantity of growth of Group A beta haemolytic streptococci was considered positive. (5)	Sensitivities: a) Culturette 36% b) Testpack 31% c) Reveal 44% d) Ventrescreen 50% e) Cards 47% Sensitivity varies with number of colonies grown on culture.	All NPTs found to have low sensitivity compared with other studies due to use of double culture as gold standard. No attempt to improve this low sensitivity. Impact analysis required.
Wright, 1987, USA, Streptococcal throat test	Culturette Brand 10 minute Group A ID test 5 Primary Care Nurses trained by the NPT manufacturer.	Primary care: one university affiliated family practice office. N=104 consecutive patients with pharyngitis.	5% sheep blood-agar cultures incubated in O2 limited, CO2 increased atmosphere for 18-24 hr at 35°c (5)	Sensitivity 38% Specificity 95% PPV 69% NPV 84%	High specificity but low sensitivity in primary care setting with non technologist operator.
Banchongaksorn 1997 Thailand Plasmodium falciparum	Parasight - F Rural health workers	1. Health Certres 2. Malaria mobile units n= 3,361 cases of suspected malaria.	Thick blood film (3)	Sensitivity 97% Specificity 99% PPV 97% NPV 99% Prevalence 29%	Accurate and simple test enables early treatment of malaria.
Hackelsberger 1998 Germany Helicobacter pylori	Helicobacter pylori Test - Boehringer Mannheim (Helisal Rapid Blood, Cortecs, UK) 1. 7 GPs 2. 5 Gastroenterologists 3. 1 Academic physician	1. Primary care 2. Community clinic 3. Hospital outpatients All undergoing investigation for dyspepsia. 1. n=38 2. n=101 3. n=64	2 antral, 2 corpus - 2 for histology, 2 for commercial urease. Positive if one or two positive. (5)	1. Sensitivity 89 (76-100), Specificity 81 (70-93) PPV 93 (88-97) NPV 58 (45-69) 2. Sensitivity 73 (63-83), Specificity 95 (86-100), PPV 98 (86-100), NPV 50 (34-66) 3. Sensitivity 89 (78-98), Specificity 67 (40-94), PPV 88 (78-98) NPV 67 (40-94)	No single reference standard used, results likely to be affected by spectrum bias. High prevalence of H.pylori in sample tested (72%) giving high PPV and low NPV. May be reversed with lower prevalence.
Talley 1998 Australia Helicobacter pylori	Helicobacter pylori Test Boehringer Mannheim (Helisal Rapid Blood, Cortecs, UK) 18 GPs	Primary care n=110 dyspeptic patients, GI bleeding and PPI use within a month excluded.	Hospital laboratory C14 Urea Breath Test. (5)	Sensitivity 59%  Specificity 90% Likelihood Ratio + 6.1,  Likelihood Ratio - 0.45 Prevalence 54%	C14 Urea breath test responds to current infection. Serology may also indicate 'past infection'. Some patients may not have a strong antibody response.
Jones 1997 UK Helicobacter pylori	Helisal Rapid Blood, Cortecs, UK GPs and Practice Nurses	Primary care n=123, dyspeptic patients not having had previous eradication therapy	Either Helisal ELISA of Helico G ELISA positive. (5)	Sensitivity 83% Specificity 78% PPV 83% NPV 78% Prevalence 67%	2 serology tests are not an 'accepted' reference standard.
Ditchburn, 1990, UK, Urine Test strip	Nephurtest plus leuco. Boeringher (Germany) General Practitioners	Primary care: 2 GPs working at a single site in rural UK. n=325 consecutive mid stream urine samples taken in response to a range of "relevant symptoms" (no standardised definition).	Bacterial culture by hospital laboratory. A pure growth with colony counts >105/ml was taken as positive. (4)	a) leucotest positive (n=229) Sensitivity 89%, Specificity 68% PPV 66%, NPV 90%. b) nitrite positive (n=266) Sensitivity 57%, Specificity 96% PPV 89%, NPV 79% c) blood positive (n=266) Sensitivity 76%, Specificity 62% PPV 55%, NPV 81%. d) leucotest and nitrite both negative or either positive (n=229): Sensitivity 91%, Specificity=67%, PPV 66%, NPV 92%	The high specificity of nitrite combined with the high sensitivity of the leucotest may be of value in ruling in/out UTI. In this study there was a prevalance of UTI of 32% and with a single site there is the possiblity of spectrum bias. In addition the gold standard used was different from that of Winkens below.
Messing 1987, USA, Urine test strip	Haematuria ³ trace on Haematuria dipstick  (make unspecified ). a) Lab technician b) General Public	a) Hospital lab: n=176 consecutive attenders of urology clinic providing urine specimens (both sexes and all ages). b) Home: n=626 men > 50 of whom 231 self tested for at least 3 months (daily for first 5 days then weekly.	a) Microscopy of spun urine for red blood cells. >2 red blood cells per high powered field considered positive. b) Urology clinic review (history, examination and investigations)  only dipstick positive patients followed up. (3)	a) Sensitivity 91%, Specificity 99%, PPV 98%, NPV 95%. b) 23/231 had haematuria at least once. Of these 4 were not evaluated clinically, 16 had identifiable cause of haematuria.	Results open to work-up bias as false negatives not identified. In addition not all patients with positive dipstick results were followed up.
Winkens, 1995, The Netherlands, Urine test strip	Nephur test plus leuco  General Practitioners	Primary care: 16 GPs in 12 practices. n=1388 urine samples from patients with symptoms of UTI (standard defeinition given). 77 samples considered contaminated, leaving 1311.	Laboratory culture: 0.03ml urine on 7% sheep blood agar/McConkey agar. Urine infection if >105/ml and 2 or less colony types. (6)	a) leucocyte esterase: Sensitivity 87%, Specificity 29% Likelihood ratio+1.2 Likelihood ratio-0.4 b) nitrite Sensitivity 66%, Specificity 75% LR+2.6, LR-0.5 c) red cells: Sensitivity 66%, Specificity 50% LR+1.3, LR-0.7	NPV of multistix GP test is at most 57% in this study with a prevalence of UTI of 69%. This multi centre study shows considerable operator variability. Note the differences in prevalence and reference standard compared with Ditchburn. The probability of spectrum bias makes comparison with other studies difficult.

The quality score was adapted from that devised by Reid et al(15) and is out of seven with one point each for: spectrum composition, analysis of pertinant sub groups, avoidance of work-up bias, avoidance of review bias, precision of results for test accuracy, presentation of indeterminate results and appropriateness of comparator/gold standard.

Table 2 Performance of anticoagulation near patient tests (NPTs)
 

Author, Year, Country, Type.	NPT Package(s), Operator.	Setting, Subjects	Performance Comparison (quality score)	Performance Results	Provisional Conclusions and areas for further research
Anderson, 1993, Canada, INR	Biotrack* Patient	Home n = 49 selected patients previously anticoagulated for at least one year (mostly with recurrent DVT), 9 excluded, providing 5–28 pairs of samples. Mean 12.1 pairs per patient.	Lab analyser: varied depending on site of "usual anticoagulation lab" for each patient. (6)	CC  r = 0.73 (95% CI 0.63-0.81) The portable monitor consistently underestimated INR at values > 3.0	Anticoagulation can be self monitored successfully in selected patients at home. Further research is required on the potential role of this NPT in a wider sample.
McCurdy, USA, 1992, INR	"Portable INR/PT monitor" Model not stated. Nurse	Office laboratory n= 143 paired blood specimens from 85 patients. 7 pairs excluded as ref INR>6.0.	Lab analyser: MLA 700, photooptical coagulation measuring device (4)	Overall: 75% INR within 0.7 units 90% INR within 0.9 units Accuracy was best at INRs of around 3.0	Systematic variations in NPT vs laboratory standard. These may not be large enough to be clinically significant. Impact analysis required.
Rose, 1993, USA, PT & APTT.	COAG-1.** Trained nurses, Lab technician.	Hospital laboratory, hospital in patients, Office practice. n = 204 (PT) and 236 (APTT) samples from unknown number of patients with both normal and disordered coagulation.	Lab analyser: Coag-A-Mate X-2 (4)	Correlation Coefficients (CC): a) PT  surgical intensive care unit r=0.75 coronary care unit r = 0.73 cardiology clinic r=0.86 b) APTT surgical intensive care unit r=0.67 coronary care unit r=0.81	Acceptable results achieved in secondary care setting. Small numbers of outliers adversely affected the results. The APTT aspects of this work are not relevant to primary care. Further work required in a primary care setting.
Seamark, 1997, UK, INR	Coagucheck Practice Nurses	Single practice n = 306 samples from 120 patients anticoagulated for a variety of conditions.	Lab analysers: IL Combinet PT FIB-HS  (4)	Systematic underestimation of INR compared with gold standard, particularly with Nycomed. Mean difference from mean INR: - 0.32  (+/-2 SD 0.55 - -1.20)	Variation probably not of clinical significance.

*Biotrack 512 and Biotrack now known as Coaguchek **COAG-I now known as TAS (CDI USA).

Table 3 Performance of near patient tests for haemoglobin, C reactive protein, and erythrocyte sedimentation rate
 

Author, Year, Country, Type.	NPT Package(s), Operator,	Setting, Subjects	Performance Comparitor (quality score)	Performance Results	Provisional Conclusions and areas for further research
Dinant, 1989, Netherlands, ESR	Unspecified ESR method(s) as previously used in general practice (1987) ESR: Sterilin holding devices with detailed instructions. (1988) Unspecified General Practice staff	Primary Care: 5 practices. Centrally prepared blood samples source not specified.	Lab analysis: Westergren's method (4)	a) CC r =0.83 b) CC r =0.97	Accurate ESR estimations can be obtained in primary care by using a standardised method. Further work required to delineate clinical value of the test.
Hjortdahl, 1991, Norway, C reactive protein.	NycoCard CRP (serum). Auxillary staff – mainly nurses	Primary care: 10 practices. Performance: samples from 288 consultations	Lab analyser: Turbidimetric assay (reagents from Orion Diagnostics, Finland applied to a Cobas Bio centrifugal analyser) (6)	CC: r = 0.85 for 195 samples where best guess of exact value was available	This NPT is moderately inaccurate in a primary care setting. Information is lacking as to whether this inaccuracy is clinically relevant. The serum test has now been superseded by a whole blood test.
Neville, 1987, UK, Haemoglobin	HemoCue. a) Lab technician, b) Nurses	a) Hospital out patient clinic b) Primary care: 3 urban practices a) 103 random selection of samples (hospital & GP) b) 235 random samples from GP practices	Lab analyser: ELT 800 WS (3) (4)	CCs: a) r=0.99 – no further data b) r=0.61- Sensitivity of abnormal test 88.5% Specificity of abnormal test 77.6%	Inaccuracy may occur if this NPT is used by non technicians in a primary care setting. Further work is required to ascertain whether this can be reduced with training and also to assess the impact of the NPT on primary care.
Søndenaa, 1992, Norway, C reactive protein	NycoCard CRP (serum). Operator not specified	Setting not specified. 158 consecutive patients with a tentative diagnosis of appendicitis.	a) Lab analyser: T-antiserum CRP b) Diagnosis of appendicitis confirmed on histology. a) (4) b) (5)	a) CC: r = 0.92 with exact agreement in 126/158 samples b) Results varied depending on cut off value of CRP chosen. For CRP>10mg/l: Sensitivity = 58% Specificity = 70%	High levels of accuracy possible with this NPT but the value of CRP in the diagnosis of acute appendicitis is questionable..

Table 4 Performance of clinical chemistry near patient tests (NPTs)
 

Author, Year, Country.	NPT Package(s), Operator, Type	Setting, Subjects	Performance Comparitor (quality score)	Performance Results	Provisional Conclusions and areas for further research
Erickson, 1993, USA, Clinical chemistry multi analyser	PCA i-STAT a) Hospital Nurses. b) Lab technician,	a) Accident and Emergency department, n= 82. b) Hospital Lab, n=142. (n=82 from above and 60 from outpatients)	a) PCA I-STAT used by Lab technician b) Laboratory analysis: Ektachem 700XRC (4)	a) CCs: Sodium r = 0.979; Potassium r =0.993 Chloride r =0.954; Urea r =0.994 Glucose r =0.987; Haematocrit r =0.974 b) All means of differences <2 c) CCs: Sodium r = 0.937; Potassium r =0.993 Chloride r =0.904; Urea r =0.996 Glucose r =0.992; Haematocrit r =0.952	Statistical analysis involved manufacturers of the NPT. Inappropriate analyses used.
Nanji, 1988, Canada, Clinical chemistry multi analyser.	BMD Reflotron, Ames Seralyser, Abbott Vision, Kodak DT-60. 14 Office staff All given half an hour of training.	Hospital Lab n= 352	The same NPT operated by a laboratory technician in a hospital laboratory. (3)	No of results differing by 10% from results obtained by hospital technician: Reflotron: 7% Seralyser: 42% Vision: 2% DT-60: 21%	NPTs with the fewest operator dependent steps performed best.
Sands, 1995, USA, Clinical chemistry multi analyser.	PCA i-STAT Research Assistant	Hospital Emergency Dept n= 960 patients requiring a blood test from the central lab while the research assistant was available.	Laboratory analysis: Ektachem 700XRC/Beckman CX3 Coulter S+IV counter NOVA statProfile-6 analyzer (5)	Mean (SD) of difference of I-STAT PCA vs central laboratory value: Sodium = -0.2 (-6 - +3) Potassium = -0.17 (-0.07 - +0.2 Chloride = 0.46 (-1.5 - +1.0) Urea = 0.36 (-1.07 - +1.78) Glucose = 0.47 (-0.66 - +1.65) Haematocrit = 4.48 (-0.5 - +7.5)	Effect of NPT assessed indirectly with inadequate control. The use of a research assistant as operator may limit the generalisability of this work.
Koch, 1987, USA, Cholesterol	BMD Reflotron, Ames Seralyser, Abbott Vision, Chrometrics Cholesterol Test System Kodak DT-60. Single technician trained by the manufacturers.	Office laboratory n= 109 volunteers and selected out patients	Blind assessment by CDC Atlanta using cholesterol reference method. (4)	CCs (serum) Vision r = 0.99 Seralyzer r = 0.96 Reflotron r = 0.97 Chrometrics r = 0.98 DT-60 r = 0.99	Analysers now replaced by more advanced technology
Phillips, 1988, Australia, Cholesterol	BMD Reflotron. Specially trained operator	Office laboratory. n=80 volunteers	Hitachi 705 analyser using cholesterol esterase/ cholesterol peroxidase monotest (Boehringer-Mannheim) (6)	Mean (SD) difference of Hitachi-Reflotron 0.22mmol/l (0.30)	The use of a specially trained operator limits the generalisability of these results. The authors suggest that the technology is suitable for use in the community but did not evaluate it in that setting.
Pope, 1993, UK, HbA1c	Ames DCA 2000 4 evaluations undertaken a) Lab technician b) Clinic medical staff c) Clinic medical staff d) Clinic nursing staff	Hospital outpatient clinics and outreach clinic Consecutive patients with attending diabetic out patients. a) n=48 (IDDM & NIDDM) b) n=19 (IDDM children) c) n=24 (Pregnant DM) d) n=15 (General NIDDM)	DIAMAT HPLC system (6)	Mean difference HbA1c (95% CI) [DCA 2000- DIAMAT] a) -0.69% (-1.42 - +0.04) b) -0.93% (-1.93 - +0.07%) c) -0.29% (-1.09 - +0.51%) d) -0.77% (-1.30 – +0.24)	Systematic under estimation of HbA1c in all cases. Moderate inaccuracy may not be clinically significant.
Stahl, 1997, Denmark, Blood glucose	1. Accutrend 2. Exactech and Medisense 3. Glucometer 4. HaemoCue 5. Hypocount 6. Reflolux GPs and nurses	Office laboratories and practices n= 265 samples	Citrated standard samples of whole blood. Reference aliquot haemolysed at point of NPT test to halt glucolysis. Tested with Unimate 7 Glucose GDH. (3)	Mean differences % +/- 1SD –15 +/- 13 n=14 –5 +/- 10 n= 19 4 +/- 14 n= 25 10 +/- 8 n= 41 5 +/- 10 n= 77	Small numbers may limit reliability of results for comparing analysers. Likely to be adequate for monitoring, but not for initial diagnosis of diabetes.
de Grauw, 1995, The Netherlands, Microalbuminuria	Micral test. GPs and practice assistants	Primary care: 10 practices. N=401 patients with NIDDM under GP care. First morning samples collected on 3 consecutive days. Samples positive for nitrite/protein excluded (n=84) leaving 317.	Immunological nephelometric measurements (Hyland-Disc 120) (4)	Standard >20mg/l Sensitivity 67% Specificity 93% PPV 74% NPV 90% LR+ 9.6 LR- 0.35	The NPT is accurate in ruling out the diagnosis if repeated samples are tested. However 21% of the patients in this sample were excluded due to macroalbuminuria.
Deviaud, 1993, France, Pregnancy test kits	27 pregnancy test kits a) Lab technician (tested all 27 kits) b) Lay women (tested only the 10 kits found to have 100% sensitivity and specificity by the technicians)	a) Hospital laboratory b) Home.	Abbot Testpack HCG urine. (5/6)	At twice the claimed detection limit: For all 27 kits: a) Lab Technicians Sensitivity 3-100% Specificity 20-100% For the 10 best performing kits: b) Lay Women Sensitivity 76-100% Specificity 0-100% Lay women had a high risk of false positive results.	Overall results from lab technicians better than lay women. This may reflect an advantage gained from training in the use of these tests.

Table 5 Evaluations of the impact of near patient tests (NPTs) conducted alongside performance evaluations
 

Author, Year, Country, Type.	NPT Package(s), Operator,	Setting, Subjects	Impact Comparison (quality not formally assessed)	Impact Results	Provisional Conclusions and areas for further research
Anderson, 1993, Canada, INR	Biotrack. Patient	Home n = 49 patients all previously anticoagulated for at leas one year (mostly with recurrent DVT), 9 excluded, providing 5-28 pairs of samples; Mean 12.1 pairs per patient.	Acceptability (questionnaire) & Harm Impact (follow up of complications: for 6-24 months (total 533 patient months)) No control group used and no blinding of subjects.	97% of patients would prefer to use portable monitor in future. Only one complication in one patient (spontaneous subarrachnoid haemorrhage; INR at time of event = 1.3)	Further research required on the potential role of this NPT in unselected patients, particularly those where inaccuracy may be greater (elderly, short term anticoagulation)
Dinant, 1989, The Netherlands, ESR	a) Unspecified ESR method(s) as previously used in general practice (1987) b) ESR: Sterilin holding devices with detailed instructions. (1988) General Practice staff	Primary Care: 5 general practice centres. n=unknown	Effect of standardised ESR method in improving performance over time.	Statistically significant changes in repeatability both between practices and within practices shown over time (1987 vs 1988). No control group used.	Further work required to delineate clinical value of the test along with any meaningful impact evaluation.
Hjortdahl, 1991, Norway, C reactive protein	NycoCard CRP (serum) Auxillary staff – mainly nurses	Primary care Samples from  a) 208 "new" consultations b) 68 "follow up" consultations.	Impact measured by  i) Responses to requests for clinical information gained ii) On a 10cm visual analogue scale (0 no impact, 10 very great impact)	i) "Clinical information gained" a) Bacterial infection ruled in/out 60%. Disease activity 41%. No info 7%. b) Bacterial infection ruled in/out 46%. Disease activity 40%. No info 16%. ii) VAS score: Mean (SD) 4.9 (2.7)	Using an uncontrolled method, little impact on patient care was found. This serum test has now been superseded by a whole blood test.
True, 1986, USA, Streptococcal throat test	Culturette Brand 10 minute Group A ID test Family physicians on research team.	Primary care (US university family practice office). n= approx 280 individuals presenting to family practice with symptoms suggestive of pharyngitis for whom throat culture was ordered.	a) Effect of NPT on  b) ordering of throat culture c) symptomatic treatment, no antibiotics d) antibiotics prescribed before test result known e) antibiotics prescribed when test result positive f) antibiotic prescription given to be filled only if culture result positive	a) Throat culture Before 98%; After 99% b) Symptomatic treatment Before 58%; After 63% c) Antibiotics before test results known Before 27%; After 9% d) Antibiotics only when test results known Before 10%; After 26% e) Antibiotic Rx only if culture positive Before 6%; After 2%	Likely to be prone to bias: Pre and post design. Retrospective pre data collection. Likelihood that data abstracters not blind to the period to which notes relate.
Pope, 1993, UK, HbA1c	Ames DCA 2000 (Bayer diagnostics Ltd, Basingstoke, UK) a) Lab technician b) Clinic medical staff c) Clinic medical staff d) Clinic nursing staff	Hospital outpatient clinic  Consecutive patients with attending diabetic out patients. a) n=48 (IDDM & NIDDM) b) n=19 (IDDM children) c) n=24 (Pregnant DM) d) n=15 (General NIDDM)	Ease of use (n=4 operators) and perceived impact (n=18 patients)	All operators found DCA 2000 easy to use; sample result time (9min) too long for busy clinics. NPT "influenced" management in 9/18 cases	Methodology of impact evaluation not well developed. More robust and larger studies are required.
Jones 1997 UK Helicobacter pylori	Helisal Rapid Blood, Cortecs, UK GPs and Practice Nurses	Primary care n=123, dyspeptic patients not having had previous eradication therapy	GPs able to use test results as they wished. 3 month follow up for changes in management, consultantations, prescribing and referrals. No control group.	67/68 H.pylori positive patients had eradication therapy. Treatment altered in 37/55 negative patients No other significant changes	An RCT of the 'test and eradicate' strategy is required.
Sands, 1995, USA, Clinical chemistry multi analyser	PCA i-STAT Research assistant	Hospital Emergency Dept n= 960 patients requiring a blood test from the central lab while the research assistant was available.	Effect of NPT on turn around time and decisions to admit, discharge and treat.	103/960 patients' admission/discharge was felt to have been altered  91/960 patients' treatment was felt to have been altered. 166/960 patients' admission/discharge/treatment was felt to have been altered. All results were physicians' opinions on the effect of at least one NPT result.	The effect of the NPT on clinical management was not assessed objectively and had no controls. This design may leave the results open to bias.
Tsai, 1994, USA, Clinical chemistry multi analyser	PCA i-STAT Research assistant	Hospital Emergency Dept n= 210 patients	Perceived effect of NPT on turn around time and decisions to admit, discharge and treat. Also estimate of effect on cost.	Mean turnaround time for NPT was 8 min vs. mean of 59 min for central laboratory. 59/210 patients' course of care was felt to have been altered in physician's opinion.	Effect of the NPT was assessed indirectly with no controls. There was no objective assessment of the clinical impact of the faster result.