Journal of Thrombosis and Thrombolysis

, Volume 30, Issue 3, pp 263–275

An evaluation of patient self-testing competency of prothrombin time for managing anticoagulation: pre-randomization results of VA Cooperative Study #481–The Home INR Study (THINRS)

Authors

    • Center for Health Services Research in Primary CareVA Medical Center
    • Division of General Internal Medicine, Department of Medicine, and Center for Clinical Health Policy ResearchDuke University Medical Center
    • Duke Clinical Research Institute
  • R. Lynne Ruybalid
    • Research and Development Service (151)Jerry L. Pettis VA Medical Center
  • Lauren Uyeda
    • Cooperative Studies Program Coordinating Center (151K)VA Palo Alto Health Care System
  • Robert G. Edson
    • Cooperative Studies Program Coordinating Center (151K)VA Palo Alto Health Care System
  • Ciaran Phibbs
    • Health Economics Resource CenterVA Palo Alto Health Care System
    • Department of Pediatrics and Center for Primary Care and Outcomes ResearchStanford University School of Medicine
  • Julia E. Vertrees
    • VA Cooperative Studies Program Clinical Research Pharmacy Coordinating Center
  • Mei-Chiung Shih
    • Cooperative Studies Program Coordinating Center (151K)VA Palo Alto Health Care System
    • Division of Biostatistics, Department of Health Research and PolicyStanford University
  • Alan K. Jacobson
    • Research and Development Service (151)Jerry L. Pettis VA Medical Center
    • Department of Internal MedicineLoma Linda University
  • David B. Matchar
    • Center for Health Services Research in Primary CareVA Medical Center
    • Division of General Internal Medicine, Department of Medicine, and Center for Clinical Health Policy ResearchDuke University Medical Center
    • Program in Health Services and Systems ResearchDuke-National University of Singapore Graduate Medical School
  • for the THINRS Site Investigators
Article

DOI: 10.1007/s11239-010-0499-8

Cite this article as:
Dolor, R.J., Ruybalid, R.L., Uyeda, L. et al. J Thromb Thrombolysis (2010) 30: 263. doi:10.1007/s11239-010-0499-8

Abstract

Prior studies suggest patient self-testing (PST) of prothrombin time (PT) can improve the quality of anticoagulation (AC) and reduce complications (e.g., bleeding and thromboembolic events). “The Home INR Study” (THINRS) compared AC management with frequent PST using a home monitoring device to high-quality AC management (HQACM) with clinic-based monitoring on major health outcomes. A key clinical and policy question is whether and which patients can successfully use such devices. We report the results of Part 1 of THINRS in which patients and caregivers were evaluated for their ability to perform PST. Study-eligible patients (n = 3643) were trained to use the home monitoring device and evaluated after 2–4 weeks for PST competency. Information about demographics, medical history, warfarin use, medications, plus measures of numeracy, literacy, cognition, dexterity, and satisfaction with AC were collected. Approximately 80% (2931 of 3643) of patients trained on PST demonstrated competency; of these, 8% (238) required caregiver assistance. Testers who were not competent to perform PST had higher numbers of practice attempts, higher cuvette wastage, and were less able to perform a fingerstick or obtain blood for the cuvette in a timely fashion. Factors associated with failure to pass PST training included increased age, previous stroke history, poor cognition, and poor manual dexterity. A majority of patients were able to perform PST. Successful home monitoring of PT with a PST device required adequate levels of cognition and manual dexterity. Training a caregiver modestly increased the proportion of patients who can perform PST.

Keywords

AnticoagulationPatient self-testingAtrial fibrillationMechanical heart valve

Introduction

Anticoagulation (AC) with warfarin reduces the risk of thromboembolism (TE) in prophylaxis of venous thrombosis after high-risk surgery, treatment of venous thrombosis and pulmonary embolism, and prevention of systemic embolism in patients with acute myocardial infarction, valvular heart disease, atrial fibrillation (AF), tissue heart valves, or mechanical heart valves (MHV). Despite compelling evidence supporting the value of AC and the importance of high quality AC, warfarin remains underused and, for those receiving treatment, dosing is often suboptimal [1]. Considering the fact that warfarin has a narrow therapeutic index, both issues are critically important, and present challenges to AC management.

Current models of AC management range from (a) the provision of care by a provider with no special education in AC, (b) management within an AC clinic, (c) patient self-testing (PST), and (d) patient self-management (PSM) subsequent to PST. PST involves use of a home–monitoring device that measures prothrombin time (PT). It is reported that PST can lead to greater quality of AC as measured by percent of days the International Normalized Ratio (INR) is within target range, when compared to care provided in an AC clinic [26]. A recent meta-analysis indicated that TE and death rates can be significantly reduced in patients with self-management [7]. However, both PST and PSM require that the patient successfully perform the test at home, which in turn implies that he/she is willing to perform the test at home, and is cognitively and physically capable of performing it in a reliable fashion after adequate training.

Current evidence regarding characteristics of patients who can successfully be trained to perform PST is scarce. In general, most trials that demonstrate superior results with PST or PSM are only able to randomize 25–43% of eligible patients [4, 6, 8, 9]. Recognizing the characteristics of patients who can and cannot be successfully trained to perform PST is important because of the time and resources involved, as well as the potential risks to patients less than properly skilled. The results of Part 1 of “The Home INR Study” (THINRS), which included screening, training, and PST competency assessment, are reported here.

Methods

The Department of Veterans Affairs (VA) Cooperative Studies Program (CSP) #481, THINRS, was a two-part 28-site randomized control trial of the clinical impact of PST. PST testing was done with the ProTime® Microcoagulation System which is FDA-approved for PST at home and is composed of the device, the cuvette containing the reagents and the proprietary finger-stick tool (Tenderlett). The design of the THINRS trial has been previously described [10]. In brief, patients with a history of MHV and/or AF who required long-term AC with warfarin were eligible. Sites were asked to pre-identify mentally competent patients on warfarin with AF or MHV using the VA’s electronic medical records and to obtain referrals from and review patient lists for the AC clinic, cardiology, cardiac surgery and related services. Sites were also encouraged to post flyers about the study and hand out letters describing the trial. In Part 1 of the trial, patients were screened, trained and tested for competency in using the INR device over a 2–4 week period. In Part 2, individuals capable of performing PST were randomized with equal allocation to either a PST arm with regularly scheduled testing, or an AC clinic management arm with testing every 4 weeks. All patients assigned to PST in Part 2 were told to test once a week except for two groups (approximately 100 patients each) which were randomly assigned to test twice a week or every 4 weeks for a substudy assessing PST frequency (conducted at six sites). THINRS patients were followed for 2–4.75 years, with the primary endpoint being time to first event among major bleed, stroke, and death.

The institutional review board for each participating VA site approved the study. All patients gave written informed consent. Trial enrollment into Part 1 took 32 months; follow-up of patients continued for an additional 25 months after enrollment ended.

Part 1 procedures

During the screening visit, study candidates were assessed for eligibility, including: (a) diagnosis of persistent or intermittent AF and/or a MHV, (b) scheduled to receive warfarin indefinitely, (c) expected to survive for the duration of the study, (d) no current intracranial bleed or other contraindication to warfarin, (e) be willing (or have a caregiver willing) to perform PST, (f) be willing to be randomized, (g) have adequate cognitive and language skills to follow the protocol and related instructions, (h) be willing to participate for the full duration of the study, (i) signed the Part 1 consent form, and (j) not be currently enrolled in another randomized clinical trial of a drug or device. From eligible patients we collected demographics, medical history, warfarin use, and medications. In addition, we collected measures of numeracy, literacy, cognition, dexterity, and satisfaction with AC. For caregivers, we collected demographics and measures of numeracy, literacy, cognition, and dexterity. Numeracy was assessed by the ability to read a series of numbers aloud (note: a validated instrument did not exist when the study was initially designed). Literacy was assessed with the REALM survey [11]. We used the short portable mental status questionnaire (SPMSQ) as a measure of cognition [12]. Dexterity was measured using an abbreviated version of the ABILHAND instrument [13], and satisfaction with care using the Duke Anticoagulation Satisfaction Survey (DASS) [14].

Training consisted of viewing a videotape supplied as part of each INR device kit, demonstration of the use of the INR device by the clinic personnel, plus practice tests and a qualifying test by the patient and/or caregiver. We also instructed patients on how to call a toll-free number to enter their INR readings and to use paper diaries to record the date, number of cuvettes used, INR results, and ER visits or hospitalizations.

The patient was instructed to use the INR device twice a week at home for 2 weeks and return for a PST competency assessment. At the competency assessment visit, the research staff downloaded information from the INR device, collected the patient diaries, and performed the same testing procedures done at the training visit by the patient/caregiver and study coordinator using two sets of devices and supplies (patient’s and clinic’s). At the site investigators’ discretion, patients who had difficulty with PST at 2 weeks could home test for an additional 2 weeks and have a second competency assessment visit. Patients who were capable of PST after 2–4 weeks of practice were eligible for randomization into Part 2 of the study.

Statistical analyses

Data analyses consisted of two-group comparisons of continuous variables (t-test or Wilcoxon rank sum test as appropriate) and categorical variables (Fisher exact test or chi-squared test as appropriate), using SAS version 9.1.3 (SAS Institute Inc, Cary, NC). We also examined the ability of a set of pre-specified independent variables to predict whether patients would pass the competency assessment via a multivariate logistic regression model, using R version 2.9.1 in conjunction with Harrell’s Design and Hmisc libraries [15]. Specifically, prior to data analysis, the investigators identified nine patient characteristics as predictors of competency status, which included linear terms for age, number of persons in household, and scores for literacy, numeracy, cognition, and dexterity, and binary terms for prior stroke (yes/no), some college education (yes/no), and married or not. A multivariate logistic model which included all predictors and no interactions was fitted and the concordance measure c [16] and R2 were calculated. Here the measure of concordance c is the area under the receiver operating characteristic (ROC) curve and is the probability that among a pair of patients, one competent and one not competent, the competent patient had the higher predicted probability of being competent. R2 is the proportion of variation in the outcome explained by the regression model and is a measure of the association between the predicted values and the outcome. We performed resampling validation to determine if the predicted probabilities of competency from the fitted model were likely to accurately predict responses for patients not used to build the model [15]. While results cited in this paper are based on the full model, we also applied fast backward elimination to identify which variables were highly predictive of competency status, and verified that this reduced model had similar but slightly lower values of c and R2.

Results

Twenty-eight VA Medical Centers screened 3745 patients; of these, 3643 received training to use the home INR device, 2931 passed the competency assessment, and 2922 were randomized to continue with Part 2 of the study (Fig. 1). Overall, 80% (2931 of 3643) of the trained patients were able to perform PST prior to randomization. The primary reason for not passing the competency assessment was failure to return for that assessment which took place 2–4 weeks after training (n = 508, 14%). Other reasons for non-randomization included: not meeting entry criteria (n = 42), not completing training (n = 60), not passing training (n = 78), and not passing the competency assessment (n = 126).
https://static-content.springer.com/image/art%3A10.1007%2Fs11239-010-0499-8/MediaObjects/11239_2010_499_Fig1_HTML.gif
Fig. 1

Participant flow diagram. Superscript alphabet (a): includes 2884 who passed both training and competency, plus 38 randomization errors (24 did not pass training but passed competency assessment; 12 passed training but not competency assessment; and two failed both)

Baseline characteristics of the 3643 patients who underwent PST training are shown in Table 1. The mean age of the study population was 67.9 years (range 23–99), and a majority were Caucasian (92%), and male (98%). Indications for chronic AC were AF (83%), MHV (23%) or both (6%). Most patients (79%) had been on warfarin for longer than a year and 8% were on AC for <3 months. Co-morbid illnesses included hypertension (71%), diabetes (34%), congestive heart failure (29%), prior stroke (10%), and prior transient ischemic attack (TIA, 10%). Very few patients had a previous history of a bleeding disorder (1%) or warfarin-related bleed (3%). Concomitant medications used by study patients that could affect bleeding or AC included aspirin (28%), Plavix (2%), amiodarone (8%), and LMW heparin (<1%). A majority of patients were functionally literate (86%) and satisfaction with AC was fairly evenly distributed between low (39%), medium (31%), and high (29%) scores.
Table 1

Baseline characteristics of patients who were trained: excludes patients who were screened and not trained (n = 102)

Characteristic

Subjects who tested alone

Subjects who tested with a caregivera

All subjects

Did not pass alone (n = 100)

Passed alone (n = 3223)

Total trained alone (n = 3323)

P-valueb

Did not pass with caregiver (n = 6)

Passed with caregiver (n = 314)

Total trained with caregiver (n = 320)

Did not pass (n = 106)

Passed (n = 3537)

Total trained (n = 3643)

P-value

Demographics

Age, mean (SD)

71 (8.6)

67.3 (9.5)

67.4 (9.5)

<0.0001

81.8 (9.8)

72.5 (8.8)

72.7 (8.9)

71.6 (9)

67.8 (9.6)

67.9 (9.6)

<0.0001

Race

 Caucasian

90 (90%)

2955 (92%)

3045 (92%)

ns

6 (100%)

289 (92%)

295 (92%)

96 (91%)

3244 (92%)

3340 (92%)

ns

 Black/African American

8 (8%)

202 (6%)

210 (6%)

ns

0 (0%)

23 (7%)

23 (7%)

8 (8%)

225 (6%)

233 (6%)

ns

 Asian

0 (0%)

11 (<1%)

11 (1%)

ns

0 (0%)

0 (0%)

0 (0%)

0 (0%)

11 (<1%)

11 (<1%)

ns

 Native Hawaiian/Pacific Islander

0 (0%)

10 (<1%)

10 (<1%)

ns

0 (0%)

0 (0%)

0 (0%)

0 (0%)

10 (<1%)

10 (<1%)

ns

 Native American/Alaskan Native

1 (1%)

57 (2%)

58 (2%)

ns

0 (0%)

1 (<1%)

1 (<1%)

1 (1%)

58 (2%)

59 (2%)

ns

 Other

3 (3%)

26 (1%)

29 (1%)

ns

0 (0%)

2 (1%)

2 (1%)

3 (3%)

28 (1%)

31 (1%)

ns

Ethnicity, Hispanic/Latino

5 (5%)

220 (7%)

225 (7%)

ns

2 (33%)

33 (11%)

35 (11%)

7 (7%)

253 (7%)

260 (7%)

ns

Gender, male

99 (99%)

3168 (98%)

3267 (98%)

ns

6 (100%)

309 (98%)

315 (98%)

105 (99%)

3477 (98%)

3582 (98%)

ns

Highest level of education completed

 High school or less

40 (40%)

1287 (40%)

1327 (40%)

ns

5 (83%)

202 (64%)

207 (65%)

45 (42%)

1489 (42%)

1534 (42%)

ns

 College, no degree

28 (28%)

1044 (32%)

1072 (32%)

 

0 (0%)

70 (22%)

70 (22%)

28 (26%)

1114 (32%)

1142 (31%)

 

 College degree or higher

32 (32%)

890 (28%)

922 (28%)

 

1 (17%)

42 (13%)

43 (13%)

33 (31%)

932 (26%)

965 (27%)

 

Marital status, married

60 (60%)

2032 (63%)

2092 (63%)

ns

3 (50%)

250 (80%)

253 (79%)

63 (59%)

2282 (65%)

2345 (64%)

ns

Number of members in household, mean (SD)

1.9 (1.1)

2 (1)

2 (1)

0.031

1.8 (0.4)

2.4 (0.9)

2.4 (0.9)

1.8 (1)

2.1 (1)

2.1 (1)

0.009

Medical history

Bleeding disorder

0 (0%)

18 (1%)

18 (1%)

ns

0 (0%)

3 (1%)

3 (1%)

0 (0%)

21 (1%)

21 (1%)

ns

Major warfarin-related bleed

3 (3%)

104 (3%)

107 (3%)

ns

0 (0%)

18 (6%)

18 (6%)

3 (3%)

122 (3%)

125 (3%)

ns

AF

75 (75%)

2679 (83%)

2754 (83%)

0.0331

6 (100%)

278 (89%)

284 (89%)

81 (76%)

2957 (84%)

3038 (83%)

0.0492

MHV

29 (29%)

733 (23%)

762 (23%)

ns

0 (0%)

64 (20%)

64 (20%)

29 (27%)

797 (23%)

826 (23%)

ns

AF and MHV

5 (5%)

194 (6%)

199 (6%)

ns

0 (0%)

28 (9%)

28 (9%)

5 (5%)

222 (6%)

227 (6%)

ns

Arrhythmia other than AF

7 (7%)

335 (10%)

342 (10%)

ns

1 (17%)

41 (13%)

42 (13%)

8 (8%)

376 (11%)

384 (11%)

ns

Congestive heart failure

24 (24%)

891 (28%)

915 (28%)

ns

2 (33%)

125 (40%)

127 (40%)

26 (25%)

1016 (29%)

1042 (29%)

ns

Angina/coronary artery disease

13 (13%)

547 (17%)

560 (17%)

ns

0 (0%)

68 (22%)

68 (21%)

13 (12%)

615 (17%)

628 (17%)

ns

Diabetes mellitus

38 (38%)

1097 (34%)

1135 (34%)

ns

1 (17%)

106 (34%)

107 (33%)

39 (37%)

1203 (34%)

1242 (34%)

ns

Hypertension

64 (64%)

2291 (71%)

2355 (71%)

ns

4 (67%)

218 (69%)

222 (69%)

68 (64%)

2509 (71%)

2577 (71%)

ns

Stroke

20 (20%)

278 (9%)

298 (9%)

<0.0001

2 (33%)

62 (20%)

64 (20%)

22 (21%)

340 (10%)

362 (10%)

0.0002

TIA

10 (10%)

309 (10%)

319 (10%)

ns

0 (0%)

46 (15%)

46 (14%)

10 (9%)

355 (10%)

365 (10%)

ns

Dementia

1 (1%)

12 (<1%)

13 (<1%)

ns

0 (0%)

13 (4%)

13 (4%)

1 (1%)

25 (1%)

26 (1%)

ns

Psychosis

2 (2%)

24 (1%)

26 (1%)

ns

0 (0%)

9 (3%)

9 (3%)

2 (2%)

33 (1%)

35 (1%)

ns

Social history

Frequent falls

6 (6%)

98 (3%)

104 (3%)

ns

0 (0%)

22 (7%)

22 (7%)

6 (6%)

120 (3%)

126 (3%)

ns

Homebound

2 (2%)

26 (1%)

28 (1%)

ns

1 (17%)

15 (5%)

16 (5%)

3 (3%)

41 (1%)

44 (1%)

ns

Nursing home

0 (0%)

8 (<1%)

8 (<1%)

ns

0 (0%)

1 (<1%)

1 (<1%)

0 (0%)

9 (<1%)

9 (<1%)

ns

Number of alcoholic drinks per week, mean (SD)

2.2 (5.6)

2.2 (5.2)

2.2 (5.2)

ns

0 (0)

2.1 (6.3)

2 (6.3)

2 (5.4)

2.2 (5.3)

2.1 (5.3)

ns

Warfarin history

AC treatment time

 <3 months

6 (6%)

252 (8%)

258 (8%)

ns

0 (0%)

23 (7%)

23 (7%)

6 (6%)

275 (8%)

281 (8%)

ns

 3–6 months

5 (5%)

198 (6%)

203 (6%)

 

0 (0%)

21 (7%)

21 (7%)

5 (5%)

219 (6%)

224 (6%)

 

 >6 months to 1 year

3 (3%)

221 (7%)

224 (7%)

 

1 (17%)

19 (6%)

20 (6%)

4 (4%)

240 (7%)

244 (7%)

 

 >1 year

86 (86%)

2551 (79%)

2637 (79%)

 

5 (83%)

251 (80%)

256 (80%)

91 (86%)

2802 (79%)

2893 (79%)

 

Average weekly warfarin dose in mg, mean (SD)

36.6 (15.6)

36.2 (15.9)

36.2 (15.9)

ns

27.4 (10)

32.3 (14.5)

32.2 (14.5)

36 (15.4)

35.8 (15.8)

35.8 (15.8)

ns

Current medications

Aspirin

22 (22%)

905 (28%)

927 (28%)

ns

0 (0%)

78 (25%)

78 (24%)

22 (21%)

983 (28%)

1005 (28%)

ns

Plavix

0 (0%)

53 (2%)

53 (2%)

ns

0 (0%)

8 (3%)

8 (3%)

0 (0%)

61 (2%)

61 (2%)

ns

Ticlid

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

0 (0%)

Amiodarone

6 (6%)

259 (8%)

265 (8%)

ns

0 (0%)

24 (8%)

24 (8%)

6 (6%)

283 (8%)

289 (8%)

ns

LMW heparin

0 (0%)

6 (<1%)

6 (<1%)

ns

0 (0%)

1 (<1%)

1 (<1%)

0 (0%)

7 (<1%)

7 (<1%)

ns

Baseline measures

Literacy (REALM); possible scores 0–66

 ≤60 (functionally illiterate)

17 (17%)

402 (12%)

419 (13%)

ns

1 (20%)

96 (33%)

97 (33%)

18 (17%)

498 (14%)

516 (14%)

ns

 >60

82 (83%)

2819 (88%)

2901 (87%)

 

4 (80%)

196 (67%)

200 (67%)

86 (83%)

3015 (86%)

3101 (86%)

 

Numeracy, mean (SD); possible scores 0–18

17.9 (0.6)

17.9 (0.5)

17.9 (0.5)

ns

17.6 (0.9)

17.4 (2.3)

17.4 (2.3)

17.8 (0.6)

17.9 (0.9)

17.9 (0.8)

ns

Cognition, mean (SD); possible scores 0–10

9.6 (1.2)

9.8 (0.5)

9.8 (0.5)

<0.0001

8.7 (2.4)

9.4 (1.3)

9.4 (1.3)

9.5 (1.3)

9.8 (0.6)

9.8 (0.6)

0.001

Dexterity, mean (SD); possible scores 0–10

9.6 (0.9)

9.8 (0.6)

9.8 (0.6)

<0.0001

7.5 (4)

9.3 (1.5)

9.3 (1.6)

9.5 (1.3)

9.8 (0.7)

9.8 (0.7)

<0.0001

Duke Anticoagulation Satisfaction Scale (DASS); possible scores 25–225

 Low: >48

42 (45%)

1227 (38%)

1269 (38%)

ns

3 (50%)

148 (47%)

151 (47%)

45 (45%)

1375 (39%)

1420 (39%)

ns

 Medium: 38–48

31 (33%)

1017 (32%)

1048 (32%)

 

3 (50%)

90 (29%)

93 (29%)

34 (34%)

1107 (31%)

1141 (31%)

 

 High: <38

21 (22%)

968 (30%)

989 (30%)

 

0 (0%)

74 (24%)

74 (23%)

21 (21%)

1042 (30%)

1063 (29%)

 

aBecause of the small number of patients with caregivers who did not pass training, results from significance testing comparing these patients to those with caregivers who passed training are not displayed

bP-values from the comparison of subjects who passed training to those who did not

Table 1 also includes summary statistics by training status (passed, did not pass) separately for patients who trained alone, those who required caregiver help, and overall. Of the 3643 patients who underwent training, 3223 (88.4%) successfully passed PST training alone, 314 (8.6%) passed with a caregiver, and 106 (2.9%) did not pass. Results of univariate significance tests for all trainees and the subset of patients who trained alone are very similar and show that younger age, larger household size, history of AF, absence of prior stroke, and better cognition and dexterity were associated with a higher likelihood of passing PST training.

Baseline characteristics of caregivers who participated in PST training are shown in Table 2. Most caregivers were female (93%), married to the study patient (70%), and on average younger than the patient (mean age 61.4 years, SD = 12).
Table 2

Baseline characteristics of caregivers that assisted patients at their last Part 1 traininga

Characteristicb

Caregivers who did not pass (n = 6)

Caregivers who passed (n = 314)

All caregivers (n = 320)

Demographics

Age, mean (SD)

70.2 (9.7)

61.2 (12)

61.4 (12)

Age range

58–81

21–84

21–84

Gender, male

0 (0%)

22 (7%)

22 (7%)

Highest education level completed

 High school or less

1 (17%)

160 (52%)

161 (52%)

 College, no degree

2 (33%)

84 (27%)

86 (28%)

 College degree or higher

3 (50%)

62 (20%)

65 (21%)

Relationship to patient

 Spouse

3 (50%)

214 (70%)

217 (70%)

 Sibling

0 (0%)

10 (3%)

10 (3%)

 Other relative

1 (17%)

54 (18%)

55 (18%)

 Non-relative friend

2 (33%)

18 (6%)

20 (6%)

 Other

0 (0%)

10 (3%)

10 (3%)

Baseline measures

Literacy (REALM); possible scores 0–66

 ≤60 (functionally illiterate)

0 (0%)

26 (9%)

26 (8%)

 >60

6 (100%)

279 (91%)

285 (92%)

Numeracy, mean (SD); possible scores 0–18

17.3 (1)

17.9 (0.3)

17.9 (0.4)

Cognition, mean (SD); possible scores 0–10

9.8 (0.4)

9.9 (0.3)

9.9 (0.3)

Dexterity, mean (SD); possible scores 0–10

9.7 (0.8)

9.8 (0.7)

9.8 (0.7)

aBecause of the small number of caregivers who did not pass training, results from significance testing are not displayed

bCharacteristics are displayed only for the first caregiver trained for each patient

Results of the PST clinic-based training (Table 3) demonstrate that the number of cuvettes used and testing attempts made during the practice phase of training were higher in the cohort that did not pass training (both P < 0.0001). There was no difference in the number of cuvettes used or testing attempts made by the study coordinator using the clinic device. For the qualifying test by the tester (patient alone or with caregiver help) using the patient’s device and cuvettes, the mean number of cuvettes used and testing attempts were higher in the cohort that did not pass training (P < 0.0001). Reasons behind their inability to pass training include trouble with performing the finger stick and/or placing blood in the cuvette within the 2-min timeframe for sample collection (both actions were essential components for obtaining an INR result independently). A quarter of the patients that failed clinic training were also unable to correctly use the toll-free number and enter their INR value.
Table 3

Clinic training results at the last Part 1 training visit

Characteristic

Did not pass training (n = 106)

Passed traininga (n = 3537)

Total (n = 3643)

P-value

Tester

 Patient alone

99 (93%)

3223 (91%)

3322 (91%)

0.0134

 Patient with help from caregiver

6 (6%)

314 (9%)

320 (9%)

 

 Missing

1 (1%)

0 (0%)

1 (<1%)

 

Practice by tester using patient device and cuvettes

 Number of cuvettes used to practice, mean (SD)

2 (1.4)

1.2 (0.8)

1.2 (0.8)

<0.0001

 Number of testing attempts made during practice, mean (SD)

2.6 (2)

1.3 (1)

1.4 (1)

<0.0001

Test by clinic staff using clinic device and cuvettes

 Number of cuvettes used to practice, mean (SD)

1.2 (0.6)

1.1 (0.4)

1.1 (0.4)

ns

 Number of testing attempts made during practice, mean (SD)

1.2 (0.6)

1.2 (0.5)

1.2 (0.5)

ns

Qualifying test by tester using patient’s device and cuvettes

 Number of cuvettes used, mean (SD)

1.9 (1.8)

1.2 (0.5)

1.2 (0.6)

<0.0001

 Number of testing attempts, mean (SD)

1.9 (1.8)

1.3 (0.6)

1.3 (0.7)

<0.0001

 Able to perform finger stick (yes)

60 (57%)

3533 (100%)

3593 (99%)

<0.0001

 Able to obtain INR independently

  Yes, little or no difficulty

31 (29%)

3285 (93%)

3316 (91%)

<0.0001

  Yes, lots of difficulty

7 (7%)

251 (7%)

258 (7%)

 

  Unable to perform test

66 (62%)

1* (<1%)

67 (2%)

 

  Missing

2 (2%)

0 (0%)

2 (<1%)

 

 INR call to 1–800 number (yes)

26 (25%)

3505 (99%)

3531 (97%)

<0.0001

aOne patient was deemed to have passed clinic training even though he was unable to obtain an INR independently due to problems with the device

The 3565 patients given an INR device to perform blood tests twice a week at home for 2 weeks included the 3537 who passed training and 28 who did not pass but were given an INR device. Of these, 3057 (86%) returned for the PST competency assessment visit. Table 4 shows the results of these PST competency assessment visits. About 88% (2698 of 3057) of those assessed passed after 2 weeks of testing, and of the 253 who failed after 2 weeks and came in for an additional 2 weeks of testing, 233 (92%) passed on the second try. Patients who failed the PST competency assessment visit had less cuvette integrity (e.g., evidence of improper storage or handling, foil packages opened, cuvette past expiration date) and higher cuvette wastage (defined as using >150% of normally required cuvettes in the 2 week period). As was the case for training, patients who had trouble performing a qualifying test failed because of an inability to perform the finger stick or obtain the INR independently.
Table 4

Self-testing assessments at the last Part 1 assessment visit

Characteristic

Did not pass assessment (n = 126)

Passed assessment (n = 2931)

Totala (n = 3057)

P-value

Tester

 Patient alone

112 (89%)

2693 (92%)

2805 (92%)

0.0013

 Patient with help from caregiver

12 (10%)

238 (8%)

250 (8%)

 

 Missing

2 (2%)

0 (0%)

2 (<1%)

 

Integrity of cuvettes (good)

116 (92%)

2835 (97%)

2951 (97%)

0.0002

Qualifying test by tester using patient’s device and cuvettes

 Able to perform finger stick (yes)

71 (56%)

2925 (100%)

2996 (98%)

<0.0001

 Able to obtain INR independently

  Yes, little or no difficulty

31 (25%)

2837 (97%)

2868 (94%)

<0.0001

  Yes, lots of difficulty

36 (29%)

93 (3%)

129 (4%)

 

  Unable to perform test

52 (41%)

1 (< 1%)

53 (2%)

 

  Missing

7 (6%)

0 (0%)

7 (<1%)

 

 Cuvette wastageb (yes)

15 (12%)

28 (1%)

43 (1%)

<0.0001

 Number of weeks of testing

   2 weeks

106 (84%)

2698 (92%)

2804 (92%)

0.0016

   4 weeks

20 (16%)

233 (8%)

253 (8%)

 

 Competent to proceed to Part 2 (yes)

0 (0%)

2931 (100%)

2931 (96%)

<0.0001

 Adverse event since last contact

0 (0%)

0 (0%)

0 (0%)

aExcludes 508 patients who passed training but did not return for the self-testing assessment

bDefined as >150% of normally required number of cuvettes

In building the multivariate logistic model to predict PST competency, we limited the analyses to the 2880 patients who did not require caregiver assistance and who either went through competency assessment or failed clinic training. Of these patients, we excluded two patients who had missing values for the characteristics of interest (both lacked a literacy score). Among the remaining 2878 patients, 2692 (94%) passed competency assessment and 186 (6%) did not. The multivariate model including all predictors and no interactions produced a moderate concordance measure c = 0.721 and a low R2 of 0.098. Three predictors were significant in the multivariate model: age, prior stroke, and dexterity score (P < 0.0001, 0.0087, and 0.0001, respectively). Estimates from resampling validation indicated that if this model was applied to a different dataset, the values of c and R2 would be 0.705 and 0.076, respectively. The majority of patients had a high predicted probability of being competent based on the fitted model (e.g., 2801 or 93% had a predicted competency probability greater than 0.8). The ROC curve in Fig. 2 indicates that the fitted model did not provide a cutoff that can be used to adequately identify patients who were able to pass the competency assessment and those who failed. For example, using a 90% predicted competency probability as the cutoff yielded 85% sensitivity and 43% specificity.
https://static-content.springer.com/image/art%3A10.1007%2Fs11239-010-0499-8/MediaObjects/11239_2010_499_Fig2_HTML.gif
Fig. 2

ROC curve showing ability to predict PST competency (yes/no) in a logistic regression model using a linear combination of these patient characteristics as predictors: age (years), prior stroke (yes/no), some college education (yes/no), married (yes/no), number of persons in household, and scores for literacy, numeracy, mental status, and dexterity. “◄” indicates the cutoff for the displayed probability of being competent based on the fitted model (e.g., if competency is defined as having a probability of at least 0.95, the values for sensitivity and 1-specificity are about 0.49 and 0.20, respectively)

The average PST training session lasted 95 ± 46 min, with 25% less than 1 h, 50% between 1 and 2 h, 20% between 2 and 3 h, and 5% longer than 3 h. Sites were allowed to train patients in groups, but most training sessions involved one patient. Under three scenarios for the level of the trainer (100% by Registered Nurse (RN); 50% each by RN and Licensed Practical Nurse (LPN); 100% by non-nurse aide), the estimated average personnel costs to conduct clinic training for a patient is $72.32, $57.83, and $35.97, respectively. Including unscheduled phone contacts and clinic visits while the patient was home testing, the average time spent per patient was 96 min. The average competency visit lasted 20 min and the mean number of competency tests was 1.08. Combining the personnel costs for training, unscheduled contacts, and the competency assessment, the per patient personnel costs under the three scenarios described above were $91.23, $72.96, and $45.37, respectively.

Discussion

We found that approximately four out of five patients who underwent training to perform PST demonstrated competency with the home INR device. Patients who were not competent to perform PST had higher numbers of practice attempts, higher cuvette wastage, and were less able to obtain an INR result independently. Factors associated with failure to perform PST included age, previous stroke history, poor cognition, and poor manual dexterity. Failure was not associated with educational level or literacy. Training a caregiver modestly increased the proportion of patients who could perform PST from 74 to 80%.

The breakdown of subjects trained in THINRS who were male (98%) and Caucasian (92%) is similar to that in the cohort of Veterans with AF (99% male and 93% Caucasian). It is worthwhile to note that the point estimates for percent of subjects who passed training on PST were similar for males (3244 of 3340; 97%) and females (293 of 303; 97%), White (3477 of 3582; 97%) and non-White (60 of 61; 98%), and for female caregivers (292 of 298; 98%).

The results of multivariate logistic regression, based on patients who tested without caregiver assistance, indicate that while age, prior stroke and dexterity score were significant predictors of competency, they in combination with six other variables only had moderate ability to distinguish competent from non-competent patients (c = 0.721) and the predicted outcomes had a low correlation with the observed competency status (R2 = 0.098).

Of the 3135 patients who either underwent competency assessment (3057) or whose tester did not pass training (78), 93.5% (2931) were deemed competent to perform home testing. Given this high success rate and the relatively low cost associated with training and follow-up during 2 weeks of practice home testing, it seems reasonable to allow any interested patient (with caregiver assistance if needed) to be screened, trained, and go through assessment for competency to do home testing. Of course, this depends on whether home testing is found to be an efficacious and cost-effective method of INR monitoring, and Part 2 of THINRS was designed to answer this and other questions.

Table 5 outlines the previous PST/PSM literature with regards to the interventions compared, number of PST/PSM training sessions, number of trained participants, and the proportion capable of PST after training [2, 46, 8, 9, 1734]. The THINRS PST competency rate of 80% immediately after training is comparable to other trials when a broad AC population is given the opportunity to try PST; trials with higher competency rates had excluded subjects with severe handicaps prior to PST training. The common reasons for inability to perform PST were visual impairment and poor manual dexterity. Some trials allowed a caregiver to assist patients with these types of barriers. Not surprisingly, the major reason for withdrawal from PST (despite successful use of the device) was the patient’s preference for venipuncture, or a lack of confidence to self-manage AC dosing. A recent Cochrane review of 14 trials on self-monitoring and self-management of AC stated that PSM and PST were not feasible in up to half of the patients requiring anticoagulant therapy [35]. This estimate was derived from the population of 11,738 that was sampled in the 14 trials, of which 7974 were either excluded from the trial or decided not to participate. On average, 24.9% (range 0 to 57.3%) were unable to complete self-monitoring or self-management. The Cochrane estimate differs from the PST competency rate, defined as the proportion of all patients offered PST who passed the training, which is calculated for THINRS and the other studies listed in Table 5. The trials involved in the Cochrane meta-analysis differed from THINRS in several respects. First, THINRS only counted patients who provided informed consent, so we did not tally refusals and those excluded by their primary care physicians prior to the consenting process. Second, the meta-analysis included trials of self-management which requires additional training over and above what was needed for THINRS. Third, THINRS participants had to be U.S. Veterans age 18 or older with AF and/or a MHV; trials in the meta-analysis included non-Veterans and patients with venous TE as a treatment indication, plus most trials were conducted in Europe.
Table 5

Overview of randomized, controlled trials investigating PST/PSM

Study, publication year

Comparison

Training description

Number trained

PST/PSM competency

Sawicki, 1999 [17]

PSM versus clinic

Three sessions

90

NR

Beyth, 2000 [2]

PST versus clinic

Two sessions

132

NR

Cromheecke, 2000 [18]

PSM versus clinic

Two sessions

50

98%

Fitzmaurice, 2002 [19]

PSM versus clinic

Two sessions

27

85%

Gadisseur, 2003, 2004 [4, 20]

PST versus PSM versus clinic

Three sessions

180

88%

Khan, 2004 [8]

PST versus clinic

One session

85

99%

Sunderji, 2004 [21]

PSM versus clinic

Two sessions

70

81%

Gardiner, 2004 [22]

PST versus clinic

Two sessions

44

89%

Voeller, 2005 [23]

PSM versus clinic

NR

101

NR

Christensen, 2006, 2007 [24, 25]

PSM versus clinic

NR

50

NR

Gardiner, 2005 [26]

PSM versus PST

NR

104

74%

Dauphin, 2008 [27]

PSM versus clinic

Three to six sessions

33

91%

Horstkotte, 1998 [28]

PSM versus clinic

NR

75

NR

Koertke, 2001, 2007 [5, 29]

PSM versus clinic

NR

783

NR

Sidhu, 2001 [30]

PSM versus clinic

Two sessions

44

80%

Fitzmaurice, 2005 [9] and Jowett, 2006 [31]

PSM versus clinic

Two sessions

327

72%

Menendez-Jandula, 2005 [6]

PSM versus clinic

Two sessions

310

97%

Siebenhofer, 2007, 2008 [32, 33]

PSM versus clinic

Four sessions

99

91%

Eitz, 2008 [34]

PSM versus clinic

NR

470

NR

NR not reported, PSM patient self-management, PST patient self-testing

In conclusion, a majority of patients and their caregivers were trained to perform PST successfully within 2–4 weeks. A regression model to distinguish competent from non-competent patients had a low correlation with the observed PST competency status. The costs of PST training are minimal if performed by a medical aide in the clinic setting. Therefore, all patients who are willing to try PST should be given the opportunity to train and use the INR device at home, and if competent, continue long-term PST for AC management.

Acknowledgements

The study was sponsored by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Clinical Sciences Research and Development Service, Cooperative Studies Program, 810 Vermont Avenue, NW, Washington, DC 20420.

Funding source

The Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Clinical Sciences Research and Development Service, Cooperative Studies Program (125), 810 Vermont Ave NW, Washington, DC 20420.

Copyright information

© Springer Science+Business Media, LLC 2010