Quality of anticoagulation with warfarin in rural Chhattisgarh, India.

Background & objectives: In most of rural India, warfarin is the only oral anticoagulant available. Among patients taking warfarin, there is a strong association between poor control of the international normalized ratio (INR) and adverse events. This study was aimed to quantify INR control in a secondary healthcare system in rural Chhattisgarh, India.
Methods: The INR data were retrospectively obtained from all patients taking warfarin during 2014-2016 at a secondary healthcare system in rural Chhattisgarh, India. Patients attending the clinic had their INR checked at the hospital laboratory and their warfarin dose adjusted by a physician on the same day. The time in therapeutic range (TTR) was calculated for patients who had at least two INR visits.
Results: The 249 patients had 2839 INR visits. Their median age was 46 yr, and the median body mass index was 17.7 kg/m[2]. They lived a median distance of 78 km (2-3 h of travel) from the hospital. The median INR was 1.7 for a target INR of 2.0-3.0 (n=221) and 2.1 for a target of 2.5-3.5 (n=28). The median TTR was 13.0 per cent, and INR was subtherapeutic 66.0 per cent of the time. Distance from the hospital was not correlated with TTR. Interpretation & conclusions: INR values were subtherapeutic two-thirds of the time, and TTR values were poor regardless of distance from the health centre. Future studies should be done to identify interventions to improve INR control.

Atrial fibrillation (AF), both valvular and non-valvular, carries an increased risk of stroke and systemic embolism, which leads to major social and economic burden. Without thromboprophylaxis, stroke risk averages four per cent/year among patients with non-valvular AF and 17-18 per cent/year among patients with valvular AF1. Anticoagulation reduces this risk by 64 per cent compared to placebo and halves the risk compared with antiplatelet therapy23. The cost of treating stroke can be ₹500,000 in the first year4, a major burden given the 1.44-1.64 million strokes diagnosed each year in India alone5.

In most of the rural India, warfarin is the only oral anticoagulant available, but there are many barriers to its optimal use. Warfarin requires frequent monitoring and adjustment to maintain a therapeutic range. Subtherapeutic values increase the risk of stroke and embolism in patients with AF6, whereas supratherapeutic values increase the risk of intracranial haemorrhage. Alternatives such as direct oral anticoagulants (DOACs) are as effective as warfarin, but these are prohibitively expensive for most rural Indians and not approved for use in patients with severe mitral stenosis or mechanical heart prostheses7.

Time in therapeutic range (TTR) is the standard method of estimating the quality of anticoagulation with warfarin8, and a goal of 70 per cent or more improves outcomes9. Although some anticoagulation clinics in high-income countries could meet this goal10, US clinics average was 55 per cent11, and outside of North America and Europe, TTR was often lower including 46.5 per cent in South Africa and 50.1 in Latin America121314. Compared to patients with a TTR of approximately >70 per cent, patients with a TTR of approximately <50 per cent have double the risk of ischaemic stroke, bleeding or death15.

Of the 45 countries that enrolled participants in the ROCKET-AF trial, India had the lowest TTR of any region, averaging 35.9 per cent14. These data were collected from an urban centre, a setting with more robust healthcare infrastructure than rural India. The present study was aimed to characterize the quality of anticoagulation with warfarin in a rural, secondary healthcare system in Chhattisgarh, India, by quantifying the TTR of warfarin patients and determining the correlation of TTR with factors such as socio-economic status, caste and distance from the hospital.

Methods

This was a retrospective observational study. De-identified data were obtained from the Jan Swasthya Sahyog (JSS) electronic medical record (EMR). All patients who received warfarin between 2014 and 2016 at JSS or its subcentres were included. Patients who lacked two or more INR values in any eight-week period were excluded from the analysis.

Demographic, clinical and laboratory data were abstracted from the EMR for all patients. Indication for warfarin was abstracted manually. Distance to the medical centre was ascertained from Google Maps. All patients who had INR measured at least twice during an eight-week period were included. A Behnk Elektronik's Thrombostat magnet-based monitor (Norderstedt, Germany) was used to measure INR values. Patients attending the clinic had their INR checked in the hospital laboratory prior to their appointments, and the results were uploaded into the EMR by the time of their appointment 1-3 h later. Warfarin was adjusted by the physician on the same day after seeing and interviewing the patient in the clinic.

Ethical approval for the study was obtained by the Ethics Committee of the Emmanuel Hospital Association, New Delhi and by the Washington University School of Medicine's IRB (#201612042).

Mean TTR was determined using the Rosendaal linear interpolation method8. Analyses were done using SAS version 9.4 (SAS Institute, Cary, NC, USA) and SPSS version 25 (IBM Corp., Armonk, NY, USA).

Results

Demographic information of the patients is shown in Table I. There were 249 patients receiving warfarin, and 162 of them had more than one INR visit (87 were excluded as they did not have at least two INR readings in any eight-week period). There was a median of nine INR visits per included patient during the three-year study period. The median age was 46 yr; 135 (54.2%) were female, and the median body mass index was 17.7 kg/m2. Most patients were form Chhattisgarh (n=175, 70.3%) and the remaining were from Madhya Pradesh. Most patients were in the scheduled tribe (ST, n=86, 34.5%) or other backward class groups (OBC, n=86, 34.5%); 44 (7.7%) were from scheduled caste group. The median distance from the main JSS hospital was 78 km, which was slightly lower (72 km) among patients with two or more INR values greater than eight week apart. Indication for warfarin is shown in Table II. Most patients (62.6%) required warfarin for AF, and 10.0 per cent of patients had mechanical valves. The median INR was 1.7 for patients whose target INR was 2.0-3.0 and 2.1 for those with a target of 2.5-3.5 (Table II). Most (n=221, 88.8%) patients had a target INR of 2.0-3.0 and 11.2 per cent (n=28) had a target of 2.5-3.5. All patients with the higher target had mechanical mitral valve replacements (MVRs). The median TTR of the entire group of patients was 13.0 per cent (IQR 0-34.8). The median time subtherapeutic was 77.5 per cent (45-100), and the median time supratherapeutic was 0.0 per cent (0-11.6). Distance showed no correlation with TTR (Spearman's rho <1%). TTR was greatest among the scheduled caste [median 26.4 (IQR 10.7-44.5)] and lowest among scheduled tribes (median 0.0, IQR 0-13.4, P<0.001). Median TTR in OBCs was 20.4 (0-35.6) and 11.3 (0-34.7) in the general category patients (Table III).

Table I

Details of patients included in the study

Demographics	n (%)	Median	IQR
Age (yr)	249	46	36-53
Height (cm)	222	155	250-162
Weight (kg)	226	47.7	39.3-54.8
BMI (kg/m2)	188	17.7	13.1-20.8
Distance (km)	249	78	33-149
Distance (km) (for patients with a TTR)	162	72	28-143

BMI, body mass index; IQR, interquartile range; TTR, time in therapeutic range

Table II

Indications for warfarin, INR targets, and median INRs in the study patients

Indication for warfarin	n (%)
AF without mechanical valve	156 (62.6)
Mechanical valve	25 (10.0)
LA or LV thrombus	12 (4.8)
DVT, PE or DVT prophylaxis	54 (21.6)
Unclear	2 (0.8)
Target INR
2.0-3.0	221 (88.8)
2.5-3.5	28 (11.2)
INR (target 2-3), median (IQR)	1.7 (1.2-2.4)
INR (target 2.5-3.5), median (IQR)	2.1 (1.6-2.8)
INRs per person, median (IQR)	9 (3-19)

AF, atrial fibrillation; LA, left atrial; LV, left ventricular; DVT, deep-vein thrombosis; PE, pulmonary embolism; INR, international normalized ratio

Table III

Time in therapeutic range

TTR	Median (%)	IQR
Median TTR	13.0	0-34.8
Time below range	77.5	45.0-100.0
Time above range	0.0	0-11.6
TTR by caste
General category	11.3	0-34.7
OBC	20.0	0-35.6
SC	26.4	10.7-44.5
ST	0.0	0-13.4
Other	2.8	2.8-37.3
TTR versus distance	Spearman’s ρ=0.009, P=0.906

SC, scheduled caste; ST, scheduled tribe; OBC, other backward class

Discussion

Our study demonstrated that the quality of anticoagulation in rural Chhattisgarh was poor. Overall, TTR was very low with a median of 13.0 per cent. The low TTR is a cause of concern because among patients taking warfarin, mortality rises with lower TTR, with the highest mortality seen in patients with <30 per cent TTR16. Several factors may have contributed to the low TTR in this rural population such as lack of infrastructure and poverty. Although we hypothesized that distance also would contribute to low TTR, there was no correlation between distance and TTR. One explanation for this negative finding is that distance may not capture the difficulty of reaching the healthcare centre in terms of mode of transport (private vehicle, public transport or on foot), lost wages, childcare or cost of travel. However, a more likely explanation may be that in an area where poor public health infrastructure is ubiquitous, being closer to an overwhelmed institution makes little difference compared to being farther away.

Poverty also might have interfered with INR control. TTR correlates with socio-economic level17, and higher TTRs have been noted in anticoagulation clinics in high-income countries10. TTR increased in countries that implemented warfarin management programmes18. However, a more sustainable, long-term solution for rural India may lie in focusing on bolstering the public health infrastructure rather than establishing warfarin management programmes alone. Targeting women may also improve impact, as women are known to have poorer INR control than men19.

The TTR was low because patients spent 77.5 per cent of their time below their target INR. The effectiveness of warfarin therapy declines exponentially with subtherapeutic INR values, both in patients with AF20 and among those with mechanical valves21. Thus, these patients with subtherapeutic TTR received only a fraction of the potential benefit of their warfarin therapy and would have high risks of stroke, which could be reduced with better INR control.

The DOAC therapy, if available, would have been an option for up to 84.2 per cent (210 of 249) of these patients based on their indication for anticoagulation. Warfarin is a fraction of the upfront cost of DOACs, but has a greater monitoring burden and risk of intracranial haemorrhage22. Although DOACs are not cost saving, but these appear cost-effective among populations that have low TTR values with warfarin23.

Patient self-testing may also be an option to improve INR control. In a US Veteran's Affairs study, INR control was improved among patients who used individual home point-of-care (POC) devices, even when they had access to a managed anticoagulation clinic24. Though individual POC devices would be very expensive for the rural poor, a POC device shared in a rural community (perhaps with the help of a community health worker) could improve laboratory and clinical outcomes.

Reducing the thrombotic tendency of mechanical valves may also lead to improved TTR. The newly designed On-X valve, for example, requires an INR of only 1.5-2.025, which was common in this study. If these valves were available for the rural poor who required a MVR, their risk of stroke might decline.

The study had several limitations. First, only 162 of the 249 patients on warfarin had more than one visit, allowing TTR to be calculated. It is unknown which of the excluded patients established care elsewhere and continued to take warfarin and who discontinued warfarin. It is possible that because of this loss to follow up, INR monitoring may actually be worse than reported. Second, some patients might have run out of warfarin for several days before they were seen in clinic, which would have exaggerated the amount of time that they were presumed to be subtherapeutic. Third, as we did not have data on income or education level, we used caste as a surrogate marker of socio-economic status. Although several studies show that health indicators vary by caste26, it would have been better to have additional socio-economic variables. Finally, we did not assess whether dietary factors contributed to the subtherapeutic INR values in this population. In a previous study, dietary vitamin K has had an inconsistent effect on the therapeutic warfarin dose27.

In conclusion, the quality of anticoagulation in rural Chhattisgarh was below standards, even among patients with access to JSS health centres. Future studies should focus on identifying potential interventions in this population, such as access to POC monitoring, genetic risk-based dosing, DOAC therapy and the overall improvement of public health infrastructure in this neglected area.