The profile and frequency of known risk factors or comorbidities for deep vein thrombosis in an urban district hospital in KwaZulu-Natal

Background Although deep vein thrombosis (DVT) is a preventable disease, it increases the morbidity and mortality in hospitalised, patients, resulting in considerable economic health impact. The identification and primary prevention of risk factors using risk assessment and stratification with subsequent anti-thrombotic prophylaxis in moderate- to severe-risk categories is the most rational means of reducing morbidity and mortality. Aim and setting The aim of the study was to describe the profile and frequency of known risk factors or comorbidities of hospitalised medical patients with ultrasound-diagnosed DVT in an urban district hospital in KwaZulu-Natal. Methods A retrospective review of clinical notes of all medical patients (age ≥ 13 years) admitted to the hospital with ultrasound-diagnosed DVT between July and December 2013. Results The median age was 40 years (interquartile range 32–60 years) and female preponderance was 72.84%. HIV and tuberculosis emerged as the prevalent risk factors, accounting for 51.85% and 35.80%, respectively. Other risk factors observed were recent hospitalisation (34.57%), smoking (25.93%), previous DVT (19.75%) and congestive cardiac failure (18.52%). Conclusion DVT in our study occurred predominantly in young female patients unlike previous studies where patients were generally older. Furthermore, HIV and tuberculosis were the two most common known risk factors or comorbidities observed. Clinicians should have a heightened awareness of venous thromboembolism in patients with either condition or where both conditions occur together and appropriate thromboprophylaxis should be administered.


Introduction
Deep vein thrombosis (DVT), with an average annual incidence of 48 per 100 000 persons, accounts for more than half of venous thromboembolic events in the United States and 52 per 100 000 persons in Australia. 1,2 In India, an overall incidence of confirmed DVT has been shown to be 17.46 per 100 000 patients with 64% occurring in non-surgical non-trauma patients. 3 Over 200 000 South Africans suffer from DVT each year, and because most DVT is occult, the true incidence is unknown. 4 According to statistics South Africa, thromboembolic disease is responsible for 20 000 deaths per annum in South Africa. 5 However, there are currently no studies describing the prevalence of DVT in KwaZulu-Natal. The economic impact of venous thromboembolism (VTE) is enormous, as shown in a systematic review conducted in high-income countries (HIC) which showed that the economic impact on the healthcare system was because of the initial VTE diagnosis and treatment as well as the costs involved in managing recurrent hospitalisations. 6 primary coagulation abnormalities (e.g. deficiency of proteins C and S, anti-thrombin III, factor V Leiden, prothrombin 20210A) have also been documented to increase the risk of VTE. 12 Recent reports suggesting increased risk of DVT amongst patients with HIV and tuberculosis (TB) point to the increasing role of these infections in the development of DVT. Given the high burden of HIV in South Africa, it is not surprising that many of the medical patients admitted with DVT are infected with HIV. Many researchers have documented the positive correlation between thrombosis and HIV infection because HIV predisposes to a hypercoagulable state, particularly those with severe immunosuppression (CD4 < 200 cells/mL). 13,14 A study conducted in 2009 at Nelson Mandela Hospital, Mthatha, showed that the prevalence of DVT was 12.5% amongst HIVpositive patients admitted to the medical wards. 15 Identification of suspected DVT cases is difficult and sometimes these are missed. 16 However, administering prophylaxis to all patients admitted to hospital is not costeffective, especially in low-and middle-income countries. 17 The high economic burden and increased morbidity and mortality associated with VTE make it important that primary prevention of DVT with risk assessment or stratification and subsequent anti-thrombotic prophylaxis in the moderate-to severe-risk category of patients is the most rational means of managing such patients.
The Padau prediction score, which assesses the risk of VTE and identifies potential high-risk hospitalised medical patients for appropriate thromboprophylaxis, makes use of clinical components to arrive at this scores. 18 A score of < 4 is categorised as low risk whilst a score of ≥ 4 is regarded as high risk. 18 Likewise, the Wells scoring system, which is a clinical prediction rule for estimating pretest probability of DVT and further subsequent diagnostic testing developed in 1997 by Wells et al., 19 has nine clinical components with a possible score range of -2-8, and this was risk stratified into three groups, namely, high risk (≥ 3 points), intermediate risk (1-2 points) and low risk (< 1 point). It was later modified in 2003 with the addition of a further component (previously documented DVT) which was observed to increase the risk of DVT, giving a possible score range of -2-9. 20 This version reduced the risk categories into two groups, namely, likely (≥ 2 points) and unlikely (< 2 points), based on the symptoms, signs and risk factors. With the emergence of HIV and its well-documented predisposition for coagulopathy, HIV is yet to be considered as an additional risk factor in any tool aimed at assessing the risk category for thromboprophylaxis administration. Tables 1 and 2 show the modified Wells and Padau prediction scoring systems, respectively, with the assigned points for each clinical feature or component. A definitive diagnosis is made by Doppler ultrasound scan for patients stratified as likely, and the most reliable indicator of the presence of a thrombus within a vein is direct visualisation of intraluminal thrombus or non-compressibility of the vein during ultrasound. 21 These operator-dependent techniques are highly accurate in the diagnosis of DVT with a weighted mean sensitivity and specificity of 97% and 94%, respectively, for proximal DVT. 22 Use of clinical signs and symptoms alone for diagnosis of DVT cannot be relied upon because they do not reliably predict which patients have DVT. 16 The sensitivity of clinical signs and symptoms alone ranges from 60% to 96% and the specificity from 20% to 72%; therefore, it is not recommended without objective non-invasive diagnosis. 16 Given the burden of HIV and TB amongst the patients seen in our study setting, we were interested in exploring the possible impact of these infections on the profile of patients presenting with DVT in our hospital. Therefore, the aim of the study was to describe the profile and frequency of risk factors of medical patients admitted with non-invasive ultrasound-diagnosed DVT.

Research method and design
This was a retrospective descriptive study done by reviewing clinical notes at an urban district hospital in KwaZulu-Natal. The study topic was selected based on the researcher's observation of an increased number of medical patients being admitted with DVT. The clinical records of all medical patients aged ≥ 13 years who were admitted to the medical wards with ultrasound-diagnosed DVT between 1 July and 31 December 2013 were retrospectively reviewed and relevant data were extracted using a pre-defined data extraction tool. 15 Patients'

Clinical feature Points
Active cancer (treatment ongoing, within 6 months, or palliative) 1 Paralysis, paresis or recent plaster immobilisation of the lower extremities 1 Recently bedridden for 3 days or more or major surgery within 12 weeks requiring general or regional anaesthesia

Risk factors or clinical features Points
Active cancer 3

Previous venous thromboembolism 3
Reduced mobility 3 Thrombophilia 3 records were identified using the ultrasound unit register and the hospital admission or discharge register. Further review of clinical records and laboratory data (accessed from the National Health Laboratory Service database) was done in order to extract relevant parameters for the study. The data extraction tool used was developed from a validated tool used in a previous study. 15 Data were captured onto Microsoft excel spreadsheet and imported into Stata version 13 for analysis. Frequency and proportional tables, mean, median, and standard deviation were generated to describe the findings.

Ethical consideration
Ethical approval was obtained from University of  19.75%), whilst other presentations like shortness of breath, cold limb and skin changes were recorded in only 9 patients (9/81, 11.11%). Table 3 shows the demographic and clinical profile of the patients.
The most common risk factors identified in this study were HIV (n = 42) and TB (n = 29), whilst recent admission (n = 28), smoking (n = 21), previous DVT (n = 16) and congestive cardiac failure (n = 15) were other risk factors seen. Risk factors like myocardial infarction, pregnancy, recent surgery, immobility (unable to perform activities of daily living for ≥ 3 days), oral contraceptive pills (OCP) and hormone replacement therapy (HRT) accounted for a negligible proportion.

Discussion
The most common associated factors in the case series were HIV and TB. Although these findings differ from the risk factors identified in studies done in HIC, it is similar to the  Considering this context-specific prevalence of HIV in South Africa and the documented strong association between HIV and thrombosis, it is not surprising that the risk of DVT would be increased in this population. There is also evidence to support the occurrence of DVT in patients with TB infection and this may be accounted for by the fact that TB is the most common opportunistic infection in HIV-infected people in Africa accounting for the synergistic effect causing an increased risk of developing DVT. 27,28,29 In addition, many of these patients with HIV and TB coinfections may have had prolonged illnesses at home that may have contributed to their immobility, further predisposing them to DVT.
It is important to note that both Padau predictive rule and Wells criteria may need to be further modified for validation within South Africa to reflect the additional risk for DVT associated with these infections because they exclude HIV and TB, which were the two most common observed known risk factors in this study population. Further studies using appropriate designs and incorporating these scoring tools, which score HIV as risk factor, are recommended to test the validity of these tools for the local South African setting.
Despite existing evidence that age has a direct correlation with the formation of DVT resulting in an exponential increase in the incidence rate with increasing age, analysis of our study showed median age of 40 years (IOR 32-60 years). 30,31 This finding is in keeping with studies done in similar setting with median age of 37 years (IQR 15-88 years). 24 Similarly, the findings of Olubanwo 15 showed that 68.6% of the DVT cases were found in those aged < 40 years. This could be explained by the fact that HIV is most prevalent in age group 15-49 years, which constituted the age group of the majority of our study sample (69.1%). 32 Recent hospitalisation, smoking, previous DVT and congestive cardiac failure were the other risk factors observed in the study. However, risk factors such as malignancy, use of OCP, HRT and obesity were not common in the study and accounted for only 1.23% -6.15%, which is consistent with the study of Levine et al. 33 36 observed smoking as a protective factor for DVT, a systematic review and meta-analysis by Cheng et al., 37 which examined the link between smoking and VTE in the general population, found that smoking was associated with an absolute risk of 24.3 cases of VTE per 100 000 person-years. They concluded that smoking is associated with a slightly increased risk of VTE. 37 The preponderance of female gender in our study population is in keeping with findings from earlier studies but inconsistent with the finding of Levine et al. 35 of male predominance of 60%. 24 40,41 The gap identified in the study will thus lend itself to a quality improvement project to improve the management of patients with DVT.

Limitations of the study
Information bias existed in the study as it was not possible to standardise the quality of the patients' clinical notes because of the retrospective nature of the study, therefore missed or improper documentation could not be completely ruled out. Furthermore, the study was done in an urban district hospital in a high prevalence of HIV and TB area, it is not possible to generalise the findings of the study.

Recommendations
• Further review of the modified Wells criteria to include HIV with a possible score of 1 and validated for use in a population with high burden of HIV infection, for example, Sub-Saharan Africa. • The authors propose that all sick, hospitalised HIV-infected patients should be considered for thromboprophylaxis, particularly if non-ambulant, and until such time as they are fully weight-bearing ambulant. • It is important that practicing clinicians ensure that all hospitalised patients with VTE should be adequately anticoagulated prior to discharge and ideally followed up at an anticoagulation clinic.

Conclusion
The review of the profile and risk factors of patients with DVT in our study setting shows predominantly young patients with female preponderance. TB and HIV are well-recognised risk factors, either alone or in conjunction. Clinicians should have a heightened awareness of VTE in patients with either condition or where both conditions occur together and appropriate thromboprophylaxis should be administered.