Quantifying the risk for patients with a risk of venous thromboembolism (VTE) persisting for a long time is still a problem as regards prophylaxis. These are typically chronic cases with reduced mobility who are cared for at home and outpatients—often elderly—entering long-term residential nursing facilities for chronic conditions. Although it has been demonstrated that nursing home residency is a risk factor for VTE [
Investigating these populations is complicated, however, by logistic difficulties for instrumental screening for bedridden patients and those with limited mobility, nursed at home or in long-term care facilities. These patients must be reached for VTE screening wherever they live.
The aim of this study was to estimate the prevalence of deep vein thrombosis (DVT) detected by systematic compression ultrasound (CUS) examination in nonacute patients confined to bed or with very reduced mobility, cared for at home or in long-term residential facilities. We also tried to estimate the impact of VTE risk factors on DVT.
The study was conducted in a population of patients nursed at home and in two long-term care facilities in the Vimercate area. Two qualified angiologists examined each patient’s proximal veins using portable US machines, and data on risk factors were recorded by a nurse unaware of the results. The study protocol was approved by the local ethics committee, and written informed consent was obtained for each patient.
The study population included all eligible patients cared for at home with national health service assistance in the Vimercate area between September and December 2007 and all eligible patients resident in the two nursing homes that had expressed willingness to take part, during the winter of 2007-2008. Inclusion criteria were age more than 18 years, inability to attend US screening for DVT in a hospital on account of markedly reduced mobility consisting of total bed rest without bathroom privileges or needing help for restroom, and written consent from the patient or the legal guardian. Patients were excluded if they suffered any acute illness at admission or if they were receiving long-term anticoagulant therapy or prophylaxis. Patients using elastic stockings were eligible.
Data were collected by independent nurses who were given instructions on definitions and data collection techniques. The data included socio-demographic details and predefined risk factors for VTE including any past episodes of VTE, current cancer, chronic respiratory failure, chronic heart failure, neurodegenerative disorders, previous paralytic stroke that remained symptomatic, and reduced mobility, specifying the cause, how long ago it had started (months of immobilization), and the degree of immobility, distinguishing totally bedridden and sedentary patients from those who could still use the bathroom.
The number of hours spent in bed was recorded, and a note was made of whether the patient had edema in one or both legs.
All the examinations were done by two experienced angiologists using a portable US machine. Venous compression ultrasonography (CUS) was done as described previously [
Logistic regression was used to estimate the impact of VTE risk factors on the probability and subsequent occurrence of DVT. Relative risks were estimated adopting a logarithmic link function. Nonlinear effects were evaluated using restricted cubic spline functions. Putative risk factors were investigated by model selection based on information criteria (AIC). The all possible regression strategy was adopted, considering the best model with a maximum of four risk factors starting from the ten considered [
As an easy-to-use tool to compute the probability of VTE, graphically summarizing the estimated model, a nomogram was obtained starting from the logistic model results. The nomogram is particularly informative on the impact of the different risk factors. The value of each risk factor is translated into a score which is directly connected to the probability of VTE and the most important risk factors are those that give the highest scores.
Of the 251 patients considered, 221 were eligible for the study. Three were considered ineligible because they refused to give informed consent, 12 were taking anticoagulants, and 15 were still too mobile for admission. Seventy patients were nursed at home and 151 were residents in the nursing homes. Table
Main characteristics of patients. The minimum and maximum, mean, median and first (Q1) and third (Q3) quartiles are reported for continuous variables.
Continuous variables
Variable | Min–Max | Mean | Median | Q1–Q3 |
---|---|---|---|---|
Age (years) | 23–102 | 85.67 | 88 | 82–92 |
Hours/day spent in wheelchair or chair | 0–4 | 8 | 7.217 | 6–8 |
Hours/day spent in bed | 10–24 | 16.78 | 16 | 16–18 |
Time of onset of reduced mobility (months ago) | 0–1042 | 38.28 | 18 | 6–36 |
Categorical variables
Variable | No. of patients | % |
---|---|---|
Cause of limited mobility | ||
Cognitive impairment/dementia | 98 | 44.3 |
Neurodegenerative disease/ paralytic stroke | 47 | 21.3 |
Osteoarticular disease | 48 | 21.7 |
Other | 28 | 12.7 |
COPD | ||
No | 190 | 86.0 |
Yes | 31 | 14.0 |
Chronic heart failure | ||
No | 84 | 38.0 |
Yes | 137 | 62.0 |
Cancer | ||
No | 212 | 95.9 |
Yes | 9 | 4.1 |
Cerebrovascular disease | ||
No | 177 | 80.1 |
Yes | 44 | 19.9 |
Previous VTE | ||
No | 196 | 88.7 |
Yes | 25 | 11.3 |
Bathroom with help | ||
No | 128 | 57.9 |
Yes | 93 | 42.1 |
Walking with help | ||
No | 173 | 78.3 |
Yes | 48 | 21.7 |
Leg edema | ||
No | 160 | 72.4 |
Yes | 61 | 27.6 |
Bilateral | 55 | 90.2 |
Unilateral | 6 | 9.8 |
None of the patients were using elastic stockings for prophylaxis.
The CUS examination detected proximal DVT in 40 patients (18%; 95% CI, 13–24%). Three had bilateral thrombosis. There were no cases of symptomatic venous thrombosis or pulmonary embolism.
Table
Unadjusted analysis: results of the logistic model with log link to obtain relative risks for each risk factor. For each categorical variable the reference category is indicated by a relative risk of 1 and the corresponding risk of DVT is reported in brackets.
Variable | Relative risk | 95% Confidence interval | |
---|---|---|---|
Hours/day spent in bed | 1.031 | 0.940–1.132 | .511 |
Time of onset of reduced mobility | 1.001 | 1.000–1.002 | .025 |
Age | 1.007 | 0.980–1.035 | .622 |
Cardiovascular disease° | |||
No | 1 (0.190) | 0.123–0.296 | |
Yes versus No | 0.920 | 0.519–1.628 | .074 |
Long-term residential care | 1 (0.238) | 0.179–0.317 | |
Nursed at home versus long-term care | 0.240 | 0.089–0.647 | .005 |
Bathroom with help | |||
No | 1(0.203) | 0.144–0.286 | |
Yes versus No | 0.741 | 0.410–1.340 | .321 |
Previous VTE | |||
No | 1(0.173) | 0.128–0.235 | |
Yes versus No | 1.384 | 0.646–2.963 | .403 |
Causes of reduced mobility | |||
Cognitive impairment/dementia | 1(0.255) | 0.182–0.358 | <.001 |
Other versus cognitive impairment/dementia | 0.478 | 0.267–0.856 | .013 |
COPD | |||
No | 1 (0.179) | 0.132–0.243 | |
Yes versus No | 1.082 | 0.496–2.361 | .844 |
Leg edema | |||
No | 1 (0.189) | 0.137–0.260 | |
Yes versus No | 0.855 | 0.445–1.642 | .638 |
°Chronic heart failure and previous non-disabling stroke taken together. Previous disabling stroke is considered in the group “causes of reduced motility.”
The risk of DVT rose by about 1 per 1000 for each month extra in the “time of onset of reduced mobility”. The risk of DVT for patients nursed at home was 24% of that of patients in a long-term care facility. The risk of DVT for patients with causes of reduced mobility other than cognitive impairment was about half the risk for patients with cognitive impairment/dementia (Table
The model selected through AIC with at most four variables included exactly four variables, namely: Long-term residential care, Previous VTE, Time of onset of reduced mobility, and Causes of reduced mobility. This model was also the one most selected out of 200 bootstrap samples (17.5%). The second most selected model (7.5%) included Long-term residential care, Previous VTE, Bathroom with help, and Causes of reduced mobility. Long-term residential care (97%), Previous VTE (70%), Time of onset of reduced mobility (42%), and Causes of reduced mobility (64%) were the variables most selected in bootstrap samples. Table
Adjusted analysis: results of the logistic model with log link to obtain relative risks for each risk factor.
Relative risk | 95% confidence interval | ||
---|---|---|---|
Time of onset of reduced mobility | 1.001 | 1.000–1.003 | .018 |
Long-term residential care | |||
Nursed at home versus long-term care | 0.257 | 0.095–0.695 | .007 |
Previous VTE | |||
Yes versus No | 2.454 | 1.203–5.006 | .014 |
Causes of reduced mobility | |||
Other versus cognitive impairment/ dementia | 0.544 | 0.297–0.996 | .048 |
The risk factors had a significant impact. As regards the discriminant ability of the model, the area under the ROC curve was 0.709 with a correction for optimism of about 0.021, resulting in a corrected index of 0.688. The risk factors had slightly more impact in the adjusted model than the unadjusted ones. This was particularly evident for previous VTE.
The nomogram in Figure
Nomogram to estimate the probability of DVT.
The nomogram also shows that the item with the highest impact among the categorical risk factors is nursing at home versus long-term residential care. Being in a nursing home accumulates 85 points out of a maximum of about 275.
Data on the frequency of VTE among nonacute patients nursed at home or in long-term care residential homes are still scarce. Most reports refer mainly to elderly cases, generally with reduced or no mobility, who cannot easily be screened instrumentally for asymptomatic DVT, unless a sonographer goes to their bedside for the examination, as was done in the present study. However, in a study conducted ten years ago, the incidence of symptomatic VTE based on the Kansas state database for retrospective cohorts of patients nursed at home was 1.30 events per 100 person-years of observation [
Our estimate of prevalence is the first to date and the 18% of proximal DVT we found is a very high frequency which has never been described before in nonacutely ill patients who in fact, did not have DVT triggering events or so-called exposing risk factors. This elderly study population has an accumulation of predisposing risk factors, though in none of the persons screened VTE was a concern. The percentage we report is similar to that described by Bosson et al. [
Similarly to what has been reported for patients nursed at home [
Another important point is that cognitive impairment and dementia as a cause of the reduced mobility contributed more to the risk of DVT than other causes such as neurologic paralysis or chronic osteoarticular problems. No easy physiopathological explanation comes to mind, but generally in these patients who are quite different from those with acute illness, the risk factors for VTE seem very different too.
Since the overall DVT risk profile seems so different and cannot easily be assessed by instrumental screening in these patients, we thought that clinicians who manage these cases might find it useful to have a tool like a nomogram that they could consult to identify those at highest risk who could then be kept under closer surveillance, with all the necessary mechanical or pharmacological measures for prevention.
One of the main limits of this study is that we could not distinguish recent forms of DVT from older ones where the vessel involved had not been recanalized. Edema and previous VTE did not differ significantly in patients with and without DVT, so these clinical indicators cannot be used to distinguish recent or earlier forms. Therefore, the unexpectedly high prevalence of proximal DVT we observed might simply reflect the earlier development of a thrombotic disorder that had arisen far in advance of the low-mobility status (as a consequence of surgical interventions, hormonal treatment, and so on). The clinical consequences of this are hard to predict, and unfortunately we are unable to provide information on how many patients were treated with prophylactic or therapeutic anticoagulants after the diagnosis was communicated to the treating physician.
Another limitation is that the diagnosis of asymptomatic DVT was based on sonography, not on the standard method, which is venography. Apart from the limited accuracy of portable US devices, sonography has very limited sensitivity in identifying venous thrombosis in asymptomatic post-surgical cases [
A further potential limit is that not all the patients examined had been nursed at home or in long-term residential care for the same time. The time of onset of reduced mobility and its severity that we recorded can therefore only be considered a proxy of this information; as we cannot exclude the possibility that this might in fact influence the risk of DVT. The fact that we did not record it may limit the external validity of our prevalence figure. Finally, the model that establishes the most reliable profile of the patient at risk for DVT, and the nomogram we propose, need to be validated in a prospective study on a similar population.
The risk of proximal DVT in patients nursed at home or in long-term care facilities is a neglected issue, but probably not negligible. Some risk factors such as previous VTE, and especially cognitive impairment as the cause of reduced mobility, or long-term residential care rather than at home, could predispose these subjects to VTE. The prognostic impact of even asymptomatic DVT in nonsurgical patients is important in terms of increased mortality [
The management of the VTE risk in nonacute patients nursed at home or in long-term care still poses challenging questions, and ad hoc clinical trials are needed to answer them.
The authors thank Dr. Roberto Altese who served as one of the two sonographer angiologists in this paper, Vincenzo Abate who is the nursing coordinator for patients nursed at home in the