Risk factors and HIV infection among patients diagnosed with deep vein thrombosis at a regional/tertiary hospital in Kimberley, South Africa

Background: Common risk factors for developing deep vein thrombosis (DVT) are prolonged immobilisation, surgery, leg trauma or fracture, pregnancy, hormone therapy, heart failure, cancer, obesity, age, and smoking. The incidence of DVT in HIV-infected patients shows a two- to tenfold increase in comparison with a healthy population of similar age. This retrospective descriptive study evaluated the presence of risk factors for DVT and the prevalence of HIV infection in patients with confirmed DVT treated at Kimberley Hospital Complex, a regional/tertiary hospital in the Northern Cape Province, South Africa, from January 2010 to December 2014. Methods: Patients were identified from the ultrasound register from the Radiology Department, and all adult patients with Doppler-proven DVT during the five-year period were included in the study. The data were extracted from the patient files and captured on data sheets. Results: Of the 852 patients included in the study, most were female (n = 536, 62.9%). The median age was 45 years (range 5–94 years, interquartile range 34–58 years). More than half (n = 443, 52.0%) of the patients were HIV-positive, 333 (39.1%) HIV-negative, while 76 (8.9%) did not have a known HIV status. The most common association with DVT was HIV infection (52.0%) followed by tuberculosis (12.4%), cancer (10.9%), and smoking (9.0%). The least common risk factors were long-distance travel (0.4%) and thrombophilia (0.4%). The left popliteal vein was the most frequently affected site (42.0%) followed by left superficial femoral vein (40.8%) and left common femoral vein (34.7%). The location of the DVT was not associated with the HIV status of the patients. Conclusion: At Kimberley Hospital Complex, the most common risk factor in patients presenting with DVT was HIV infection, with more than half of the patients being HIV-positive. Other risk factors included tuberculosis, cancer and smoking.


Introduction
Venous thrombosis is the process of a clot or thrombus formation within the veins. This predominantly occurs in the leg vessels resulting in deep vein thrombosis (DVT), or in the lungs resulting in a pulmonary embolus. This is collectively referred to as venous thromboembolism (VTE) and is associated with recurrent nonfatal VTE, post-thrombotic syndrome and fatal pulmonary embolism. 1,2 Common risk factors for the development of DVT are prolonged immobilisation, surgery, leg trauma or fracture, plaster cast, paralysis or paresis, pregnancy or puerperium, hormone therapy, heart failure, long-distance travel, cancer, obesity, age, previous VTE, and smoking. 3 Copur et al. reported that the human immunodeficiency virus (HIV) also increases the risk of VTE. 4 In a retrospective review of patients with diagnosed DVT in an urban district in KwaZulu-Natal, South Africa, Awolesi et al. reported that HIV (51.85%) and tuberculosis (TB) (35.80%) were prevalent risk factors. 5 Additional risk factors also reported in their study population were recent hospitalisation, smoking, previous DVT, and congestive heart failure.
The incidence of DVT in HIV-infected patients shows a two-to tenfold increase in comparison with a healthy population of similar age. 6 The increased risk of venous thrombotic disease in HIV-infected patients could be explained by the presence of a hypercoagulable state, characterised by an increase in procoagulant factors and micro-particles, and a decrease in anticoagulant factors, including anti-thrombin III, heparin cofactor II, and the protein C pathway. 6,[7][8][9] HIV is a major public health concern worldwide. It is estimated that in 2015, 36.7 million people globally were living with HIV, 2.1 million became newly infected and 1.1 million had died of HIVrelated illnesses. 10 According to the 2012 South African National HIV Prevalence, Incidence and Behaviour Survey, the total number of people in South Africa living with HIV in 2012 was approximately 6.4 million. 11 An estimated 15.9% of South African adults aged 15 to 49 years are HIV-positive.
Data from studies done in the United States support the hypothesis that HIV-infected patients are more prone to clinically detected thromboembolic disease as compared with non-HIVinfected patients. Malek et al. examined a possible link between HIV infection and the presence of both pulmonary embolism and DVT. They reported that HIV-positive patients had statistically significant greater odds of 43% of developing a pulmonary embolism, 10% of developing DVT, and 40% of developing DVT or a pulmonary embolism. 12 Copur et al. reported that HIVpositive patients under the age of 50 were at increased risk for VTE compared with HIV-negative patients in the same age range. 4 Studies have been done within the Southern Africa context to determine the association between HIV infection and DVT. 8,9,11,13 A retrospective review of patients with confirmed DVT at an urban Cape Town public hospital reported a clear association between HIV, TB and the development of DVT. 13 Some 64% of the patients were HIV-positive, 55.5% had TB, and 43.3% were positive for both HIV and TB.
Kimberley Hospital Complex (KHC) is a regional/tertiary hospital and the main referral centre in the Northern Cape. All patients presenting at KHC are triaged according to the Cape Triage score. 14 Patients with suspected DVT are further assessed according to the Wells score system, which categorises the probability of DVT as high, moderate or low. 15 Patients with a moderate to high probability of DVT are sent for a duplex Doppler ultrasound, while the D-dimer blood test is used to exclude DVT in low-risk patients. 16

Aim and objectives
The aim of this study was to determine the presence of DVT risk factors and the prevalence of HIV infection in patients presenting to KHC who were diagnosed with DVT using a Doppler ultrasound. A further objective was to compare the demographics and risk factors of the HIV-positive and HIV-negative patients.

Study design, sample and population
This was a retrospective descriptive study with an analytic component. The sample included all patients who had undergone a Doppler ultrasound from January 2010 to December 2014 at the Department of Radiology, KHC, for suspected DVT. Patients with findings confirming DVT were included in the analysis.

Data collection
The ultrasound register at the Radiology Department was used to identify patients with suspected DVT. Files of patients with confirmed DVT were provided by the hospital's record department. Information was captured on a data sheet and included gender, age, HIV status, site(s) of DVT, risk factors and Wells score. The Wells score is used to determine the probability of DVT according to relevant clinical findings. A score of 3-8 points is indicative of a high probability of DVT, 1-2 points of a moderate probability and -2-0 of a low probability. 15 If the HIV status results were not recorded in the patient's file, the main researcher attempted to locate the information from laboratory records.

Pilot study
A pilot study was conducted on 30 patients with confirmed DVT seen at KHC outside the study period to verify the feasibility of the study. Minor corrections were made to the data sheet. Data obtained from the pilot study were not included in the main study.

Data analysis
Data were analysed by the Department of Biostatistics, Faculty of Health Sciences of the University of the Free State (UFS). Descriptive statistics consisted of frequencies and percentages (categorical variables) and means and standard deviations or medians and percentiles (numerical variables, depending on data distribution). HIV-positive and -negative patients were compared using chi-square (categorical variables) and t-tests. An ANOVA was used to test the differences in age by HIV status.

Ethical and legal considerations
The study was approved by the Ethics Committees of the Faculty of Health Sciences, UFS (ECUFS 09/2015) and Northern Cape Provincial Department of Health (NCPHREC NC2015/007). Permission to access the ultrasound records was obtained from the Head, Department of Radiology and to access patient files from the Medical Director, KHC. Each patient was assigned a unique patient identifier to maintain confidentiality.

Demographic information
From January 2010 to December 2014, 927 patients were diagnosed with DVT according to their Doppler ultrasound results. Of these patients, 852 (91.9%) patient files were available for data analysis.
The patients' Wells scores ranged between 0 and 6, with a mean of 2.8 and a median of 3 (a score ≥ 3 points is indicative of a high probability of DVT). Patients who were HIV-positive or HIVnegative both had mean scores of 2.8. Patients with unknown HIV status had a mean score of 2.9. Of the HIV-positive patients, 290 (65.5%) had a score of ≥ 3 compared with 225 (67.6%) of the HIV-negative patients ( Figure 1).

Location of DVT
All DVTs were present in the lower limbs of the patients. It was possible for patients to have DVT at multiple sites. Table 1 indicates the number of affected sites among patients diagnosed with DVT according to their HIV status with chi-square test range: 0-9). There was no significant association between HIV status and location of the DVT site.

Risk factors
Risk factors noted in this study population included immobility (> 7 days), cancer, smoking, TB, previous DVT, pregnancy/ puerperium, heart failure, obesity (BMI > 30), recent surgery (< 6 months), trauma/fracture/plaster cast, stroke/paralysis/paresis, long-distance travel (> 8 h), the use of contraception, and thrombophilia.   Patients could have more than one risk factor.  In Figure 3, the age distribution according to HIV status is displayed.

Discussion
In this study, 62.9% of the patients with DVT were female compared with 37.1% who were male patients. Awolesi et al. had a similar gender distribution with a majority of females in their sample, based on hospital admissions. 5 Their study, however, showed a slightly younger female population affected by DVT. The highest percentage of HIV-positive female patients with DVT were in their early thirties while HIV-positive males were in their late thirties.
HIV-negative patients with DVT were mostly in their sixties.
It is not clear whether the preponderance of females in the sample is due to hormonal or gender factors, or is a reflection of differing health service usage by the genders. Statistics South Africa reported that more women attend health facilities than men to seek help in the early stage of their illness. The men believed that their illness was not serious enough to warrant care. 17 The findings of this study and those of Awolesi et al. are surprisingly similar in the prevalence of HIV in the DVT-proven populations. 5 The prevalence of HIV in the DVT patients was much higher than the prevalence of the 'general admitted population' at KHC and the general Northern Cape population. According to the 2012 South African National HIV Prevalence, Incidence and Behaviour Survey, the prevalence of HIV in South Africa is 12.2%. 11 This prevalence varies among provinces, with KwaZulu-Natal the highest at 16.9% and the Western Cape and Northern Cape the lowest at 5.0% and 7.4%, respectively.
An audit by Louw et al. at a hospital in Johannesburg reported an HIV prevalence of 84% in patients presenting with DVT. 9 Olubanwo reported a prevalence of 79.4% in a retrospective study on all DVT cases admitted to a hospital in Mthatha over a one-year period. 18 Our results as well as the literature show that HIV prevalence in DVT patients is higher compared with the prevalence of HIV in the general population. 6 The high prevalence of HIV among known DVT patients may well indicate that the relative risk of developing DVT when being HIV-positive is much higher than in the HIVnegative group, as shown by Govender et al. 19

HIV prevalence
The yearly percentage of patients with unknown HIV status among the patients with DVT varied from 18.4% in 2010 to 3.9% in 2013. Of patients with known HIV status, the HIV prevalence ranged between 52.9% in 2014 and 64.7% in 2010 ( Figure 2).
In both the male and female patients, at least 50% were HIVpositive ( Table 3). As presented in Table 4

Limitations
As this was a retrospective study, data of patients with confirmed DVT by Doppler ultrasound were obtained from the ultrasound register and patient files, and blood results from the laboratory database. This study depended on the quality of the data recorded. There were files without mention of risk factors and several without the patient's HIV results.

Conclusion
This study found a prevalence of HIV infection in more than half of patients with DVT admitted to KHC over a five-year period. Over the same period, the reported KHC prevalence of patients with HIV ranged between 7.0% and 10.9%. The data also showed Previous studies have also shown the association between HIV and VTE (DVT). 4,16 The HIV infection being a prothrombotic condition, multiple factors have been described as of HIV-related venous thromboembolism aetiology. 4,12,20 It was recognised 20 years ago that HIV disease causes a prothrombotic condition but little evidence is known regarding the mechanism. 21 A study by Jong et al. showed that initiation of antiretroviral therapy improves the markers of endothelial cell activation, coagulation and natural anticoagulation in South African HIV-infected individuals. 22 A similar study done in Denmark confirmed the result and demonstrated the improvement in cardiovascular risk. 23 There has been some decline in HIV prevalence in patients with DVT at KHC during the five-year study period, dropping from 64.7% in 2010 to 52.9% in 2014. However, this is still high compared with the prevalence of HIV among patients admitted to KHC during the same period, which ranged between 7.0% and 10.9%. 24 Considering the Wells score as measured in this study, the mean Wells scores for HIV-positive, HIV-negative and HIV status unknown patients were similar and indicative of a high probability of DVT. This is not surprising, as the sample consisted only of confirmed DVT cases. It is, in fact, more troublesome that close on a third of the Wells scores for confirmed DVT cases were below 3, which is the cut-off value for a high probability of DVT. However, most of those cases below 3 were either 1 or 2, which indicates a moderate likelihood of DVT, and only one case had a score of zero. It has been demonstrated by Oudega et al. that the Wells score might be inadequate in the primary care setting, although in a tertiary hospital such as KHC this should be less of a problem. 25 In this study, however, the selection of confirmed DVT cases (with their attendant high Wells scores) does not preclude the idea of considering HIV infection as an independent risk factor for DVTs.
For all of the DVT sites, no statistical difference was found. All of the patients in this study had DVT in their lower limbs, which confirms findings in the literature. 26,27 Labropoulos et al. also reported that lower extremity DVT is more prevalent in hospitalised patients, and is associated with significant shortand long-term morbidity and mortality. 27 Some studies give prominence to the left lower limb as a DVT site. Ouriel et al. reported that DVT was more common on the left side, and speculated that it was due to left common iliac vein