Venous thromboembolism (VTE) is a major source of morbidity and mortality for hospitalized patients, with as many as 16% of high‐risk medical patients developing VTE during their hospital stay.1, 2 Pharmacologic prophylaxis with subcutaneous heparin reduces the risk of VTE by approximately 50%,3, 4 and guidelines produced by the American College of Chest Physicians (ACCP) recommend thromboprophylaxis for patients at moderate‐to‐high risk of VTE with either low‐molecular‐weight heparin (LMWH) or unfractionated heparin (UFH).2 UFH is less expensive per dose, but meta‐analyses have suggested that UFH may be either less effective than LMWH3 or more likely to cause complications, such as bleeding5 or heparin‐induced thrombocytopenia (HIT).6 Others have argued that the efficacy and risk of bleeding with UFH and LMWH are similar.7, 8 In either case, there are few head‐to‐head studies of LMWH and UFH in medical patients and they tend to be small. In the most recent meta‐analysis, which included fewer than 4500 patients, several different low‐molecular‐weight heparins were assessed together, and the observed rate of deep vein thrombosis (DVT) with UFH was high (5.4%), with evidence suggesting publication bias.3

Given the current Joint Commission requirement9 that all medical patients either receive VTE prophylaxis or have documented a reason not to, the implications related to choosing one form of VTE prophylaxis over another are substantial on a national scale. In order to compare the effectiveness of UFH and LMWH in routine practice among hospitalized medical patients, we conducted a retrospective cohort study in a national sample of hospitals and compared the risk of VTE, bleeding, HIT, and death associated with each treatment.

METHODS

RESULTS

Our final sample included 32,104 patients who received prophylaxis at 333 hospitals (see Supporting Information, e‐Figure, in the online version of this article). Patient characteristics appear in Table 1. Most patients (66%) were over age 65; 59% were female and 61% were white. The most common primary diagnoses were pneumonia (40%) and congestive heart failure (25%). Additional risk factors for thromboembolism included cancer (13%), paralysis (8%), or diabetes (35%). Most patients' attending physicians were either internists (61%) or family practitioners (14%). Almost half of the patients were cared for at hospitals in the South (46%).

Fifty‐five percent of patients received LMWH and the remainder received UFH; 1274 (4%) patients switched type of heparin during their stay. The proportion of patients receiving LMWH at an individual hospital varied from 0% to 100% with a u‐shaped distribution, with almost one‐third of hospitals prescribing one treatment or the other exclusively (Figure 1). Similarly, the proportion of an individual physician's patients who received prophylaxis with UFH (vs LMWH) varied from 0% to 100% (Figure 1), with 51% prescribing LMWH exclusively and 31% prescribing UFH exclusively. Compared to patients who received UFH, patients receiving LMWH were older and were more likely to be white, female, and to have pneumonia. By far the biggest difference between the groups was the hospitals at which they received their care (see Supporting Information, e‐Table, in the online version of this article). Patients receiving LMWH were much more likely to be from smaller, rural, non‐teaching hospitals in the South or the West. There were also numerous small differences in comorbidities and individual VTE risk factors between the 2 groups. The only large difference was that patients with a secondary diagnosis of renal failure (for which LMWH is not US Food and Drug Administration [FDA] approved) were almost twice as likely to receive UFH.

Figure 1

We identified 163 (0.51%) episodes of VTE (Table 2). Compared to patients receiving UFH, those receiving standard LMWH had similar unadjusted rates of VTE (0.53% vs 0.48%; P = 0.54), major bleeding (0.77% vs 0.76%; P = 0.88), thrombocytopenia (1.9% vs 2.0%; P = 0.48), definite HIT (n = 1 vs n = 3; P = 0.34), and mortality (2.8% vs 3.1%; P = 0.07). Definite complications of prophylaxis (HIT or major bleed combined with the discontinuation of heparin) were more common among patients receiving UFH (0.2% vs 0.1%; P = 0.022). Patients treated with UFH had longer unadjusted lengths of stay (P < 0.0001) and higher unadjusted costs (P < 0.0001).

A propensity model for UFH treatment based upon patient characteristics and treatments was not strongly predictive of treatment (c = 0.58) and propensity matching failed to balance many of the patient characteristics. However, hospital alone, ignoring patient characteristics was strongly predictive (c = 0.91) of treatment.

In a model adjusting only for clustering within hospitals, patients treated with UFH had an odds ratio (OR) for VTE of 1.08 (95% confidence interval [CI] 0.79 to 1.49) compared to patients receiving LMWH (Figure 2). Adjustment for propensity for UFH and other covariates attenuated the effect of LMWH (OR 1.04, 95% CI 0.76 to 1.43). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH use was associated with a nonsignificant change in the odds of VTE (OR 1.14, 95% CI 0.72 to 1.81).

Figure 2

Adjusted for clustering within hospital only, patients treated with UFH had an odds ratio for major bleed of 1.38 (95% CI 1.00 to 1.91) compared to patients receiving LMWH (Figure 3). Adjustment for propensity for UFH and other covariates gave similar results (OR 1.34, 95% CI 0.97 to 1.84). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with a nonsignificant increase in the odds of major bleed (OR 1.64, 95% CI 0.50 to 5.33). When we compared the rate of transfusion across hospitals, including 576,231 additional patients who were excluded from the original analyses because they did not receive daily prophylaxis or had a diagnosis of myocardial infarction or COPD, there was a slight negative correlation between transfusion rates and use of UFH (Spearman Correlation Coefficient 0.03; P = 0.61). Hospitals that used primarily UFH had a transfusion rate of 0.60% versus 0.76% at hospitals using primarily LMWH (P = 0.54), indicating that the increased risk of major bleeding associated with UFH was not confounded by local transfusion practices.

Figure 3

Adjusted for clustering only, patients treated with UFH had an odds ratio for definite complication of 2.35 (95% CI 1.17 to 4.72) compared to those treated with LMWH. Adjustment for propensity and covariates accentuated the association (OR 2.84, 95% CI 1.43 to 5.66). When assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with an increase in the risk of definite complication (OR 2.79, 95% CI 1.00 to 7.74).

Adjusted for clustering only, patients treated with UFH had higher costs than those treated with LMWH (cost ratio 1.07, 95% CI 1.05 to 1.09). Adjustment for propensity for UFH and other covariates attenuated the association (cost ratio 1.02, 95% CI 1.00 to 1.03). Finally, when individual patients were assigned a probability of initial treatment with UFH equal to the hospital rate where they received care, treatment with UFH was associated with a nonsignificant change in the relative cost (cost ratio 0.97, 95% CI 0.90 to 1.05).

DISCUSSION

In this retrospective cohort study, we found that low‐molecular‐weight heparin and unfractionated heparin were associated with similar rates of VTE in moderate‐to‐high risk medical patients. However, unfractionated heparin was associated with a small, but higher risk of complications, even after adjustment. There were no statistical differences in rates of heparin‐induced thrombocytopenia, but this complication was exceedingly rare. Finally, overall costs associated with both treatments were similar.

A number of industry‐funded studies have compared LMWH to UFH in randomized clinical trials. These trials have generally been small and used endpoints of uncertain significance, such as asymptomatic deep vein thrombosis assessed by ultrasound. At least 3 meta‐analyses of these trials have been published. Each used different inclusion criteria. The only one to find an efficacy benefit to LMWH over UFH was heavily influenced by the inclusion of a number of studies of stroke patients.3 In that study, LMWH reduced VTE by approximately one‐third relative to UFH. The other 2 analyses found smaller reductions in DVT and pulmonary embolism (PE), and these results were not statistically significant.5, 8 Similarly, 1 analysis5 found a reduction in major bleeding events with LMWH versus UFH, whereas the other 2 studies found smaller reductions which were not statistically significant. The assessment of major bleeding is further complicated by differences in the definition of major bleeding across studies. Using a standard definition of 2 units of packed red blood cells transfused in 1 day to denote major bleeding, we found an associated reduction in bleeding with LMWH that was similar to that observed in the meta‐analyses. Moreover, patients receiving UFH were twice as likely to have a complication that resulted in stopping the prophylaxis, although these complications were overall quite rare. Lastly, there are no cost comparisons based on randomized trials. Several comparisons based on modeling have favored LMWH. One assumed that 3% of patients receiving UFH would develop HIT;13 something we did not observe. At least 3 additional analyses,1416 all funded by the manufacturer of enoxaparin, assumed that LMWH was both more effective and safer than UFH. We found that adjusted costs were similar or slightly lower with UFH than LMWH.

Our study has a number of limitations. First, its observational design makes it vulnerable to selection bias. We attempted to overcome this with rigorous multivariable adjustment, including the propensity for treatment and by using an adaptation of the instrumental variable approach. This method is of particular interest because individual hospitals were strongly predictive of choice of heparin. Still, we cannot exclude the possibility of residual confounding, especially if other outcomes, such as transfusion decisions, were also tied to specific hospital practices. Second, our study used administrative data, and therefore we could not directly adjust for certain differences which may exist between patients who received LMWH and those who received UFH. However, we did adjust for many classic risk factors for VTE. More importantly, it seems that the chance of being treated with a particular form of heparin depends more on the hospital where one receives care than on any combination of patient characteristics. Thus, apart from renal failure, for which we adjusted, it seems unlikely that there were major differences in unmeasured physiological confounders. Third, we limited our analysis to patients who received standard dosing of either type of heparin. We did this to bolster the validity of our findings, but they may not apply to unconventional dosing often observed in clinical practice. Fourth, we measured only outcomes that occurred in the hospital or that prompted a return to the hospital. VTEs which were diagnosed and treated in ambulatory care were not included. While this may have led us to underestimate the true risk of VTE, we have little reason to believe that the choice of whether to admit a patient with VTE is influenced by the original choice of VTE prophylaxis. Finally, our study was conducted before the introduction of generic LMWH, which would be expected to reduce costs associated with LMWH prophylaxis.

VTE prophylaxis for medical patients has emerged as a major focus for quality improvement initiatives. As a result, a significant proportion of general medical patients receive some form of chemoprophylaxis during their hospital stay. Small differences in efficacy or safety of different forms of prophylaxis multiplied by millions of admissions each year can have profound effects on the health of hospitalized patients. Similarly, differences in cost could also have a substantial impact on the healthcare system. We found no difference in efficacy or cost, but treatment with LMWH was less likely to be associated with subsequent transfusion of 2 or more units of packed red blood cells, a surrogate marker for bleeding. In addition, LMWH is more convenient since it can be dosed once daily, and for that reason may be more acceptable to patients. For these reasons, LMWH may be the drug of choice for inpatient prophylaxis of general medical patients. In situations where the cost of the medication itself is important, UFH represents an equally effective alternative.

Venous thromboembolism (VTE) is a major source of morbidity and mortality for hospitalized patients, with as many as 16% of high‐risk medical patients developing VTE during their hospital stay.1, 2 Pharmacologic prophylaxis with subcutaneous heparin reduces the risk of VTE by approximately 50%,3, 4 and guidelines produced by the American College of Chest Physicians (ACCP) recommend thromboprophylaxis for patients at moderate‐to‐high risk of VTE with either low‐molecular‐weight heparin (LMWH) or unfractionated heparin (UFH).2 UFH is less expensive per dose, but meta‐analyses have suggested that UFH may be either less effective than LMWH3 or more likely to cause complications, such as bleeding5 or heparin‐induced thrombocytopenia (HIT).6 Others have argued that the efficacy and risk of bleeding with UFH and LMWH are similar.7, 8 In either case, there are few head‐to‐head studies of LMWH and UFH in medical patients and they tend to be small. In the most recent meta‐analysis, which included fewer than 4500 patients, several different low‐molecular‐weight heparins were assessed together, and the observed rate of deep vein thrombosis (DVT) with UFH was high (5.4%), with evidence suggesting publication bias.3

Given the current Joint Commission requirement9 that all medical patients either receive VTE prophylaxis or have documented a reason not to, the implications related to choosing one form of VTE prophylaxis over another are substantial on a national scale. In order to compare the effectiveness of UFH and LMWH in routine practice among hospitalized medical patients, we conducted a retrospective cohort study in a national sample of hospitals and compared the risk of VTE, bleeding, HIT, and death associated with each treatment.

METHODS

RESULTS

Our final sample included 32,104 patients who received prophylaxis at 333 hospitals (see Supporting Information, e‐Figure, in the online version of this article). Patient characteristics appear in Table 1. Most patients (66%) were over age 65; 59% were female and 61% were white. The most common primary diagnoses were pneumonia (40%) and congestive heart failure (25%). Additional risk factors for thromboembolism included cancer (13%), paralysis (8%), or diabetes (35%). Most patients' attending physicians were either internists (61%) or family practitioners (14%). Almost half of the patients were cared for at hospitals in the South (46%).

Fifty‐five percent of patients received LMWH and the remainder received UFH; 1274 (4%) patients switched type of heparin during their stay. The proportion of patients receiving LMWH at an individual hospital varied from 0% to 100% with a u‐shaped distribution, with almost one‐third of hospitals prescribing one treatment or the other exclusively (Figure 1). Similarly, the proportion of an individual physician's patients who received prophylaxis with UFH (vs LMWH) varied from 0% to 100% (Figure 1), with 51% prescribing LMWH exclusively and 31% prescribing UFH exclusively. Compared to patients who received UFH, patients receiving LMWH were older and were more likely to be white, female, and to have pneumonia. By far the biggest difference between the groups was the hospitals at which they received their care (see Supporting Information, e‐Table, in the online version of this article). Patients receiving LMWH were much more likely to be from smaller, rural, non‐teaching hospitals in the South or the West. There were also numerous small differences in comorbidities and individual VTE risk factors between the 2 groups. The only large difference was that patients with a secondary diagnosis of renal failure (for which LMWH is not US Food and Drug Administration [FDA] approved) were almost twice as likely to receive UFH.

Figure 1

We identified 163 (0.51%) episodes of VTE (Table 2). Compared to patients receiving UFH, those receiving standard LMWH had similar unadjusted rates of VTE (0.53% vs 0.48%; P = 0.54), major bleeding (0.77% vs 0.76%; P = 0.88), thrombocytopenia (1.9% vs 2.0%; P = 0.48), definite HIT (n = 1 vs n = 3; P = 0.34), and mortality (2.8% vs 3.1%; P = 0.07). Definite complications of prophylaxis (HIT or major bleed combined with the discontinuation of heparin) were more common among patients receiving UFH (0.2% vs 0.1%; P = 0.022). Patients treated with UFH had longer unadjusted lengths of stay (P < 0.0001) and higher unadjusted costs (P < 0.0001).

A propensity model for UFH treatment based upon patient characteristics and treatments was not strongly predictive of treatment (c = 0.58) and propensity matching failed to balance many of the patient characteristics. However, hospital alone, ignoring patient characteristics was strongly predictive (c = 0.91) of treatment.

In a model adjusting only for clustering within hospitals, patients treated with UFH had an odds ratio (OR) for VTE of 1.08 (95% confidence interval [CI] 0.79 to 1.49) compared to patients receiving LMWH (Figure 2). Adjustment for propensity for UFH and other covariates attenuated the effect of LMWH (OR 1.04, 95% CI 0.76 to 1.43). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH use was associated with a nonsignificant change in the odds of VTE (OR 1.14, 95% CI 0.72 to 1.81).

Figure 2

Adjusted for clustering within hospital only, patients treated with UFH had an odds ratio for major bleed of 1.38 (95% CI 1.00 to 1.91) compared to patients receiving LMWH (Figure 3). Adjustment for propensity for UFH and other covariates gave similar results (OR 1.34, 95% CI 0.97 to 1.84). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with a nonsignificant increase in the odds of major bleed (OR 1.64, 95% CI 0.50 to 5.33). When we compared the rate of transfusion across hospitals, including 576,231 additional patients who were excluded from the original analyses because they did not receive daily prophylaxis or had a diagnosis of myocardial infarction or COPD, there was a slight negative correlation between transfusion rates and use of UFH (Spearman Correlation Coefficient 0.03; P = 0.61). Hospitals that used primarily UFH had a transfusion rate of 0.60% versus 0.76% at hospitals using primarily LMWH (P = 0.54), indicating that the increased risk of major bleeding associated with UFH was not confounded by local transfusion practices.

Figure 3

Adjusted for clustering only, patients treated with UFH had an odds ratio for definite complication of 2.35 (95% CI 1.17 to 4.72) compared to those treated with LMWH. Adjustment for propensity and covariates accentuated the association (OR 2.84, 95% CI 1.43 to 5.66). When assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with an increase in the risk of definite complication (OR 2.79, 95% CI 1.00 to 7.74).

Adjusted for clustering only, patients treated with UFH had higher costs than those treated with LMWH (cost ratio 1.07, 95% CI 1.05 to 1.09). Adjustment for propensity for UFH and other covariates attenuated the association (cost ratio 1.02, 95% CI 1.00 to 1.03). Finally, when individual patients were assigned a probability of initial treatment with UFH equal to the hospital rate where they received care, treatment with UFH was associated with a nonsignificant change in the relative cost (cost ratio 0.97, 95% CI 0.90 to 1.05).

DISCUSSION

In this retrospective cohort study, we found that low‐molecular‐weight heparin and unfractionated heparin were associated with similar rates of VTE in moderate‐to‐high risk medical patients. However, unfractionated heparin was associated with a small, but higher risk of complications, even after adjustment. There were no statistical differences in rates of heparin‐induced thrombocytopenia, but this complication was exceedingly rare. Finally, overall costs associated with both treatments were similar.

A number of industry‐funded studies have compared LMWH to UFH in randomized clinical trials. These trials have generally been small and used endpoints of uncertain significance, such as asymptomatic deep vein thrombosis assessed by ultrasound. At least 3 meta‐analyses of these trials have been published. Each used different inclusion criteria. The only one to find an efficacy benefit to LMWH over UFH was heavily influenced by the inclusion of a number of studies of stroke patients.3 In that study, LMWH reduced VTE by approximately one‐third relative to UFH. The other 2 analyses found smaller reductions in DVT and pulmonary embolism (PE), and these results were not statistically significant.5, 8 Similarly, 1 analysis5 found a reduction in major bleeding events with LMWH versus UFH, whereas the other 2 studies found smaller reductions which were not statistically significant. The assessment of major bleeding is further complicated by differences in the definition of major bleeding across studies. Using a standard definition of 2 units of packed red blood cells transfused in 1 day to denote major bleeding, we found an associated reduction in bleeding with LMWH that was similar to that observed in the meta‐analyses. Moreover, patients receiving UFH were twice as likely to have a complication that resulted in stopping the prophylaxis, although these complications were overall quite rare. Lastly, there are no cost comparisons based on randomized trials. Several comparisons based on modeling have favored LMWH. One assumed that 3% of patients receiving UFH would develop HIT;13 something we did not observe. At least 3 additional analyses,1416 all funded by the manufacturer of enoxaparin, assumed that LMWH was both more effective and safer than UFH. We found that adjusted costs were similar or slightly lower with UFH than LMWH.

Our study has a number of limitations. First, its observational design makes it vulnerable to selection bias. We attempted to overcome this with rigorous multivariable adjustment, including the propensity for treatment and by using an adaptation of the instrumental variable approach. This method is of particular interest because individual hospitals were strongly predictive of choice of heparin. Still, we cannot exclude the possibility of residual confounding, especially if other outcomes, such as transfusion decisions, were also tied to specific hospital practices. Second, our study used administrative data, and therefore we could not directly adjust for certain differences which may exist between patients who received LMWH and those who received UFH. However, we did adjust for many classic risk factors for VTE. More importantly, it seems that the chance of being treated with a particular form of heparin depends more on the hospital where one receives care than on any combination of patient characteristics. Thus, apart from renal failure, for which we adjusted, it seems unlikely that there were major differences in unmeasured physiological confounders. Third, we limited our analysis to patients who received standard dosing of either type of heparin. We did this to bolster the validity of our findings, but they may not apply to unconventional dosing often observed in clinical practice. Fourth, we measured only outcomes that occurred in the hospital or that prompted a return to the hospital. VTEs which were diagnosed and treated in ambulatory care were not included. While this may have led us to underestimate the true risk of VTE, we have little reason to believe that the choice of whether to admit a patient with VTE is influenced by the original choice of VTE prophylaxis. Finally, our study was conducted before the introduction of generic LMWH, which would be expected to reduce costs associated with LMWH prophylaxis.

VTE prophylaxis for medical patients has emerged as a major focus for quality improvement initiatives. As a result, a significant proportion of general medical patients receive some form of chemoprophylaxis during their hospital stay. Small differences in efficacy or safety of different forms of prophylaxis multiplied by millions of admissions each year can have profound effects on the health of hospitalized patients. Similarly, differences in cost could also have a substantial impact on the healthcare system. We found no difference in efficacy or cost, but treatment with LMWH was less likely to be associated with subsequent transfusion of 2 or more units of packed red blood cells, a surrogate marker for bleeding. In addition, LMWH is more convenient since it can be dosed once daily, and for that reason may be more acceptable to patients. For these reasons, LMWH may be the drug of choice for inpatient prophylaxis of general medical patients. In situations where the cost of the medication itself is important, UFH represents an equally effective alternative.

Venous thromboembolism (VTE) is a major source of morbidity and mortality for hospitalized patients, with as many as 16% of high‐risk medical patients developing VTE during their hospital stay.1, 2 Pharmacologic prophylaxis with subcutaneous heparin reduces the risk of VTE by approximately 50%,3, 4 and guidelines produced by the American College of Chest Physicians (ACCP) recommend thromboprophylaxis for patients at moderate‐to‐high risk of VTE with either low‐molecular‐weight heparin (LMWH) or unfractionated heparin (UFH).2 UFH is less expensive per dose, but meta‐analyses have suggested that UFH may be either less effective than LMWH3 or more likely to cause complications, such as bleeding5 or heparin‐induced thrombocytopenia (HIT).6 Others have argued that the efficacy and risk of bleeding with UFH and LMWH are similar.7, 8 In either case, there are few head‐to‐head studies of LMWH and UFH in medical patients and they tend to be small. In the most recent meta‐analysis, which included fewer than 4500 patients, several different low‐molecular‐weight heparins were assessed together, and the observed rate of deep vein thrombosis (DVT) with UFH was high (5.4%), with evidence suggesting publication bias.3

Given the current Joint Commission requirement9 that all medical patients either receive VTE prophylaxis or have documented a reason not to, the implications related to choosing one form of VTE prophylaxis over another are substantial on a national scale. In order to compare the effectiveness of UFH and LMWH in routine practice among hospitalized medical patients, we conducted a retrospective cohort study in a national sample of hospitals and compared the risk of VTE, bleeding, HIT, and death associated with each treatment.

METHODS

RESULTS

Our final sample included 32,104 patients who received prophylaxis at 333 hospitals (see Supporting Information, e‐Figure, in the online version of this article). Patient characteristics appear in Table 1. Most patients (66%) were over age 65; 59% were female and 61% were white. The most common primary diagnoses were pneumonia (40%) and congestive heart failure (25%). Additional risk factors for thromboembolism included cancer (13%), paralysis (8%), or diabetes (35%). Most patients' attending physicians were either internists (61%) or family practitioners (14%). Almost half of the patients were cared for at hospitals in the South (46%).

Fifty‐five percent of patients received LMWH and the remainder received UFH; 1274 (4%) patients switched type of heparin during their stay. The proportion of patients receiving LMWH at an individual hospital varied from 0% to 100% with a u‐shaped distribution, with almost one‐third of hospitals prescribing one treatment or the other exclusively (Figure 1). Similarly, the proportion of an individual physician's patients who received prophylaxis with UFH (vs LMWH) varied from 0% to 100% (Figure 1), with 51% prescribing LMWH exclusively and 31% prescribing UFH exclusively. Compared to patients who received UFH, patients receiving LMWH were older and were more likely to be white, female, and to have pneumonia. By far the biggest difference between the groups was the hospitals at which they received their care (see Supporting Information, e‐Table, in the online version of this article). Patients receiving LMWH were much more likely to be from smaller, rural, non‐teaching hospitals in the South or the West. There were also numerous small differences in comorbidities and individual VTE risk factors between the 2 groups. The only large difference was that patients with a secondary diagnosis of renal failure (for which LMWH is not US Food and Drug Administration [FDA] approved) were almost twice as likely to receive UFH.

Figure 1

We identified 163 (0.51%) episodes of VTE (Table 2). Compared to patients receiving UFH, those receiving standard LMWH had similar unadjusted rates of VTE (0.53% vs 0.48%; P = 0.54), major bleeding (0.77% vs 0.76%; P = 0.88), thrombocytopenia (1.9% vs 2.0%; P = 0.48), definite HIT (n = 1 vs n = 3; P = 0.34), and mortality (2.8% vs 3.1%; P = 0.07). Definite complications of prophylaxis (HIT or major bleed combined with the discontinuation of heparin) were more common among patients receiving UFH (0.2% vs 0.1%; P = 0.022). Patients treated with UFH had longer unadjusted lengths of stay (P < 0.0001) and higher unadjusted costs (P < 0.0001).

A propensity model for UFH treatment based upon patient characteristics and treatments was not strongly predictive of treatment (c = 0.58) and propensity matching failed to balance many of the patient characteristics. However, hospital alone, ignoring patient characteristics was strongly predictive (c = 0.91) of treatment.

In a model adjusting only for clustering within hospitals, patients treated with UFH had an odds ratio (OR) for VTE of 1.08 (95% confidence interval [CI] 0.79 to 1.49) compared to patients receiving LMWH (Figure 2). Adjustment for propensity for UFH and other covariates attenuated the effect of LMWH (OR 1.04, 95% CI 0.76 to 1.43). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH use was associated with a nonsignificant change in the odds of VTE (OR 1.14, 95% CI 0.72 to 1.81).

Figure 2

Adjusted for clustering within hospital only, patients treated with UFH had an odds ratio for major bleed of 1.38 (95% CI 1.00 to 1.91) compared to patients receiving LMWH (Figure 3). Adjustment for propensity for UFH and other covariates gave similar results (OR 1.34, 95% CI 0.97 to 1.84). When individual patients were assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with a nonsignificant increase in the odds of major bleed (OR 1.64, 95% CI 0.50 to 5.33). When we compared the rate of transfusion across hospitals, including 576,231 additional patients who were excluded from the original analyses because they did not receive daily prophylaxis or had a diagnosis of myocardial infarction or COPD, there was a slight negative correlation between transfusion rates and use of UFH (Spearman Correlation Coefficient 0.03; P = 0.61). Hospitals that used primarily UFH had a transfusion rate of 0.60% versus 0.76% at hospitals using primarily LMWH (P = 0.54), indicating that the increased risk of major bleeding associated with UFH was not confounded by local transfusion practices.

Figure 3

Adjusted for clustering only, patients treated with UFH had an odds ratio for definite complication of 2.35 (95% CI 1.17 to 4.72) compared to those treated with LMWH. Adjustment for propensity and covariates accentuated the association (OR 2.84, 95% CI 1.43 to 5.66). When assigned a probability of treatment with UFH equal to the hospital rate where they received care, UFH treatment was associated with an increase in the risk of definite complication (OR 2.79, 95% CI 1.00 to 7.74).

Adjusted for clustering only, patients treated with UFH had higher costs than those treated with LMWH (cost ratio 1.07, 95% CI 1.05 to 1.09). Adjustment for propensity for UFH and other covariates attenuated the association (cost ratio 1.02, 95% CI 1.00 to 1.03). Finally, when individual patients were assigned a probability of initial treatment with UFH equal to the hospital rate where they received care, treatment with UFH was associated with a nonsignificant change in the relative cost (cost ratio 0.97, 95% CI 0.90 to 1.05).

DISCUSSION

In this retrospective cohort study, we found that low‐molecular‐weight heparin and unfractionated heparin were associated with similar rates of VTE in moderate‐to‐high risk medical patients. However, unfractionated heparin was associated with a small, but higher risk of complications, even after adjustment. There were no statistical differences in rates of heparin‐induced thrombocytopenia, but this complication was exceedingly rare. Finally, overall costs associated with both treatments were similar.

A number of industry‐funded studies have compared LMWH to UFH in randomized clinical trials. These trials have generally been small and used endpoints of uncertain significance, such as asymptomatic deep vein thrombosis assessed by ultrasound. At least 3 meta‐analyses of these trials have been published. Each used different inclusion criteria. The only one to find an efficacy benefit to LMWH over UFH was heavily influenced by the inclusion of a number of studies of stroke patients.3 In that study, LMWH reduced VTE by approximately one‐third relative to UFH. The other 2 analyses found smaller reductions in DVT and pulmonary embolism (PE), and these results were not statistically significant.5, 8 Similarly, 1 analysis5 found a reduction in major bleeding events with LMWH versus UFH, whereas the other 2 studies found smaller reductions which were not statistically significant. The assessment of major bleeding is further complicated by differences in the definition of major bleeding across studies. Using a standard definition of 2 units of packed red blood cells transfused in 1 day to denote major bleeding, we found an associated reduction in bleeding with LMWH that was similar to that observed in the meta‐analyses. Moreover, patients receiving UFH were twice as likely to have a complication that resulted in stopping the prophylaxis, although these complications were overall quite rare. Lastly, there are no cost comparisons based on randomized trials. Several comparisons based on modeling have favored LMWH. One assumed that 3% of patients receiving UFH would develop HIT;13 something we did not observe. At least 3 additional analyses,1416 all funded by the manufacturer of enoxaparin, assumed that LMWH was both more effective and safer than UFH. We found that adjusted costs were similar or slightly lower with UFH than LMWH.

Our study has a number of limitations. First, its observational design makes it vulnerable to selection bias. We attempted to overcome this with rigorous multivariable adjustment, including the propensity for treatment and by using an adaptation of the instrumental variable approach. This method is of particular interest because individual hospitals were strongly predictive of choice of heparin. Still, we cannot exclude the possibility of residual confounding, especially if other outcomes, such as transfusion decisions, were also tied to specific hospital practices. Second, our study used administrative data, and therefore we could not directly adjust for certain differences which may exist between patients who received LMWH and those who received UFH. However, we did adjust for many classic risk factors for VTE. More importantly, it seems that the chance of being treated with a particular form of heparin depends more on the hospital where one receives care than on any combination of patient characteristics. Thus, apart from renal failure, for which we adjusted, it seems unlikely that there were major differences in unmeasured physiological confounders. Third, we limited our analysis to patients who received standard dosing of either type of heparin. We did this to bolster the validity of our findings, but they may not apply to unconventional dosing often observed in clinical practice. Fourth, we measured only outcomes that occurred in the hospital or that prompted a return to the hospital. VTEs which were diagnosed and treated in ambulatory care were not included. While this may have led us to underestimate the true risk of VTE, we have little reason to believe that the choice of whether to admit a patient with VTE is influenced by the original choice of VTE prophylaxis. Finally, our study was conducted before the introduction of generic LMWH, which would be expected to reduce costs associated with LMWH prophylaxis.

VTE prophylaxis for medical patients has emerged as a major focus for quality improvement initiatives. As a result, a significant proportion of general medical patients receive some form of chemoprophylaxis during their hospital stay. Small differences in efficacy or safety of different forms of prophylaxis multiplied by millions of admissions each year can have profound effects on the health of hospitalized patients. Similarly, differences in cost could also have a substantial impact on the healthcare system. We found no difference in efficacy or cost, but treatment with LMWH was less likely to be associated with subsequent transfusion of 2 or more units of packed red blood cells, a surrogate marker for bleeding. In addition, LMWH is more convenient since it can be dosed once daily, and for that reason may be more acceptable to patients. For these reasons, LMWH may be the drug of choice for inpatient prophylaxis of general medical patients. In situations where the cost of the medication itself is important, UFH represents an equally effective alternative.

In the outpatient setting, medication adherence (defined as percentage of prescribed medication doses taken by a patient during a specific time period) ranges between 40% and 80% for chronic conditions.1 During acute care hospitalization, changes are often made to patients' medication regimens, which can be confusing and contribute to nonadherence, medication errors, and harmful adverse events.2 Indeed, it is estimated that almost half of patients encounter a medication error after discharge, and approximately 12%17% experience an adverse drug event after returning home.36 It is likely that some of these adverse events may be the result of medication nonadherence.7 Improved patientprovider communication, systems to reconcile prehospitalization and posthospitalization medications, as well as development of mechanisms to enhance adherence, may prevent many of these errors and have become new targets for quality improvement.4, 8 Although postdischarge medication adherence is a crucial target for avoiding adverse events and rehospitalization, few studies have focused on understanding its incidence and predictors, in particular, patient demographic factors such as age and insurance status.911

In addition, few studies have looked at general and posthospital adherence in a population where health literacy is measured, an important area because medication changes during hospitalization may be particularly confusing for patients with low health literacy.11, 12 Health literacy is defined as the degree to which an individual has the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.13 Prior outpatient research shows that low health literacy is associated with poor patient understanding of the medication regimen and instructions for medication use, which may contribute to postdischarge medication nonadherence.14, 15 Understanding the factors associated with postdischarge medication adherence could help refine interventions that are oriented toward improving transitions in care, patient safety, and reducing unnecessary rehospitalization.

We report here on factors associated with postdischarge medication adherence using data from the Pharmacist Intervention for Low Literacy in Cardiovascular Disease (PILL‐CVD) study.16

METHODS

RESULTS

Table 1 shows descriptive baseline patient characteristics of study sample (responders) as well as nonresponders at 30 days. For the responders, the mean age of the 646 patients was 61.2 years, 94.7% were insured, and 19.3% had inadequate or marginal health literacy. Patients were prescribed an average of 8 preadmission medications. Most patients (92.3%) had a regular PCP prior to admission. Nonresponders had nonsignificant trends towards having lower health literacy, medication understanding, and baseline medication adherence.

The average postdischarge adherence score was 95% (standard deviation [SD] = 10.2%), and less than 10% of patients had an adherence score of less than 85%; overall the distribution was left‐skewed. Table 2 illustrates crude and adjusted parameter estimates for variables in the model. Table 3 shows significant findings in the fully adjusted model, which used multiple imputation techniques to account for missing data.

Intervention arm was of borderline statistical significance in predicting postdischarge adherence (P = 0.052), and so was removed from the final model. Study site, age, insurance, and baseline adherence were the only significant independent predictors of postdischarge adherence in the fully adjusted model (Table 3). For example, for every 10‐year increase in age, patients had, on average, an adjusted 1% absolute increase in their adherence score (95% confidence interval [CI] 0.4% to 2.0%). For every 1‐point increase in baseline medication adherence (based on the Morisky scale), there was a 1.6% absolute increase in medication adherence (95% CI 0.8% to 2.4%). In unadjusted analyses, patients with Medicaid were less adherent with medications after discharge than were patients with private insurance. This difference became nonsignificant in adjusted analyses, but when analyses were repeated using multiple imputation techniques, the results again became statistically significantMedicaid insurance was associated with a 4.5% absolute decrease in postdischarge adherence compared with private insurance (95% CI 7.6% to 1.4%). Study site (specifically, Brigham and Women's Hospital) was also a significant predictor of greater postdischarge medication adherence. Years of education was a significant predictor of adherence in unadjusted analyses, but was not an independent predictor when adjusted for other factors. When baseline adherence was removed from the multiple imputation model, there were no changes in which factors were significant predictors of adherence.

DISCUSSION

In this study, we found that low baseline adherence, younger age, Medicaid insurance, and study site were significant predictors of lower 30‐day medication adherence. Of particular interest is our finding regarding baseline adherence, a simple measure to obtain on hospitalized patients. It is notable that in our study, education was not an independent significant predictor of postdischarge adherence, even when baseline adherence was removed from the model. The same is true for medication understanding, cognitive function, and health literacy.

Older patients appeared more adherent with medications in the month after hospital discharge, perhaps reflecting increased interaction with the healthcare system (appointments, number of physician interactions), a greater belief in the importance of chronic medication management, or a higher level of experience with managing medications. A similar relationship between age and adherence has been shown in outpatient studies of patients with hypertension, diabetes, and other chronic diseases.2427

Medicaid patients may be less likely to remain adherent because of the plan's limited coverage of medications relative to patients' ability to pay. For example, Medicaid in Tennessee covers the first 5 generic medications at no cost to the patient but has co‐payments for additional medications and for brand name drugs. Medicaid in Massachusetts has co‐payments of $1 to $3 for each medication. Alternatively, Medicaid insurance may be a marker for other patient characteristics associated with low adherence for which we were not fully able to adjust.

Site differences were also notable in this study; these differences could have been due to differences in insurance coverage in Tennessee versus Massachusetts (which has near‐universal coverage), differences in types of insurance (eg, fewer patients at Brigham and Women's Hospital had Medicaid than at Vanderbilt), cultural and geographic differences between the 2 locations, or other differences in transitional care between the 2 sites.

This study corroborates previous literature on medication adherence (specifically unintentional nonadherence) in the outpatient setting,4, 811 for example, on the association of younger age with low adherence in certain populations. On the other hand, it may contrast with previous literature which has sometimes shown a relationship between patient education or health literacy and medication adherence.14, 15, 2835 However, previous studies have not focused on the transition from inpatient to outpatient settings. Perhaps intensive medication education in the hospital, even under usual care, mitigates the effects of these factors on postdischarge adherence. Finally, baseline adherence seems to correlate with postdischarge adherence, a finding which makes intuitive sense and has been previously reported for specific medications.36

There are several limitations to this study. Although large, the study was performed at only 2 clinical sites where most patients were white and fairly well‐educated, perhaps because patients admitted to a tertiary care center with ACS or ADHF are more affluent than general medical inpatients as a whole; this may limit generalizability. Postdischarge medication adherence might have been higher than in other patient populations given the nature of the population, possible loss‐to‐follow‐up bias, and the fact that half of the subjects received an intervention designed to improve medication management after discharge; such low rates of nonadherence in our study may have reduced our ability to detect important predictors in our models. In addition, the period of follow‐up was 30 days, thus limiting our findings to short‐term postdischarge medication adherence. Postdischarge medication adherence was based on patient self‐report, which not only assumed that the patient was still managing his/her own medications after discharge, but may also be susceptible to both recall and social acceptability bias, which might overestimate our adherence scores, again limiting our ability to detect important predictors of nonadherence. However, other studies have shown a good correlation between self‐reported medication adherence and other more objective measures,37, 38 and recall was only for 7 days, a measure used previously in the literature39, 40 and one designed to reduce recall bias. Systematic underreporting in certain patient populations is less likely but possible.

In the future, research should focus on targeting patients who have low baseline adherence to evaluate the effects of various interventions on postdischarge medication outcomes. Repeating the study in a population with a high prevalence of low health literacy might be illuminating, given that previous studies have shown that patients with low health literacy have less ability to identify their medications and have less refill adherence.29, 30

In conclusion, in patients hospitalized with cardiovascular disease, predictors of lower postdischarge adherence include younger age, Medicaid insurance, and low baseline adherence. It may be prudent to assess baseline adherence and insurance type in hospitalized patients in order to identify those who may benefit from additional assistance to improve medication adherence and medication safety during transitions in care.

In the outpatient setting, medication adherence (defined as percentage of prescribed medication doses taken by a patient during a specific time period) ranges between 40% and 80% for chronic conditions.1 During acute care hospitalization, changes are often made to patients' medication regimens, which can be confusing and contribute to nonadherence, medication errors, and harmful adverse events.2 Indeed, it is estimated that almost half of patients encounter a medication error after discharge, and approximately 12%17% experience an adverse drug event after returning home.36 It is likely that some of these adverse events may be the result of medication nonadherence.7 Improved patientprovider communication, systems to reconcile prehospitalization and posthospitalization medications, as well as development of mechanisms to enhance adherence, may prevent many of these errors and have become new targets for quality improvement.4, 8 Although postdischarge medication adherence is a crucial target for avoiding adverse events and rehospitalization, few studies have focused on understanding its incidence and predictors, in particular, patient demographic factors such as age and insurance status.911

In addition, few studies have looked at general and posthospital adherence in a population where health literacy is measured, an important area because medication changes during hospitalization may be particularly confusing for patients with low health literacy.11, 12 Health literacy is defined as the degree to which an individual has the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.13 Prior outpatient research shows that low health literacy is associated with poor patient understanding of the medication regimen and instructions for medication use, which may contribute to postdischarge medication nonadherence.14, 15 Understanding the factors associated with postdischarge medication adherence could help refine interventions that are oriented toward improving transitions in care, patient safety, and reducing unnecessary rehospitalization.

We report here on factors associated with postdischarge medication adherence using data from the Pharmacist Intervention for Low Literacy in Cardiovascular Disease (PILL‐CVD) study.16

METHODS

RESULTS

Table 1 shows descriptive baseline patient characteristics of study sample (responders) as well as nonresponders at 30 days. For the responders, the mean age of the 646 patients was 61.2 years, 94.7% were insured, and 19.3% had inadequate or marginal health literacy. Patients were prescribed an average of 8 preadmission medications. Most patients (92.3%) had a regular PCP prior to admission. Nonresponders had nonsignificant trends towards having lower health literacy, medication understanding, and baseline medication adherence.

The average postdischarge adherence score was 95% (standard deviation [SD] = 10.2%), and less than 10% of patients had an adherence score of less than 85%; overall the distribution was left‐skewed. Table 2 illustrates crude and adjusted parameter estimates for variables in the model. Table 3 shows significant findings in the fully adjusted model, which used multiple imputation techniques to account for missing data.

Intervention arm was of borderline statistical significance in predicting postdischarge adherence (P = 0.052), and so was removed from the final model. Study site, age, insurance, and baseline adherence were the only significant independent predictors of postdischarge adherence in the fully adjusted model (Table 3). For example, for every 10‐year increase in age, patients had, on average, an adjusted 1% absolute increase in their adherence score (95% confidence interval [CI] 0.4% to 2.0%). For every 1‐point increase in baseline medication adherence (based on the Morisky scale), there was a 1.6% absolute increase in medication adherence (95% CI 0.8% to 2.4%). In unadjusted analyses, patients with Medicaid were less adherent with medications after discharge than were patients with private insurance. This difference became nonsignificant in adjusted analyses, but when analyses were repeated using multiple imputation techniques, the results again became statistically significantMedicaid insurance was associated with a 4.5% absolute decrease in postdischarge adherence compared with private insurance (95% CI 7.6% to 1.4%). Study site (specifically, Brigham and Women's Hospital) was also a significant predictor of greater postdischarge medication adherence. Years of education was a significant predictor of adherence in unadjusted analyses, but was not an independent predictor when adjusted for other factors. When baseline adherence was removed from the multiple imputation model, there were no changes in which factors were significant predictors of adherence.

DISCUSSION

In this study, we found that low baseline adherence, younger age, Medicaid insurance, and study site were significant predictors of lower 30‐day medication adherence. Of particular interest is our finding regarding baseline adherence, a simple measure to obtain on hospitalized patients. It is notable that in our study, education was not an independent significant predictor of postdischarge adherence, even when baseline adherence was removed from the model. The same is true for medication understanding, cognitive function, and health literacy.

Older patients appeared more adherent with medications in the month after hospital discharge, perhaps reflecting increased interaction with the healthcare system (appointments, number of physician interactions), a greater belief in the importance of chronic medication management, or a higher level of experience with managing medications. A similar relationship between age and adherence has been shown in outpatient studies of patients with hypertension, diabetes, and other chronic diseases.2427

Medicaid patients may be less likely to remain adherent because of the plan's limited coverage of medications relative to patients' ability to pay. For example, Medicaid in Tennessee covers the first 5 generic medications at no cost to the patient but has co‐payments for additional medications and for brand name drugs. Medicaid in Massachusetts has co‐payments of $1 to $3 for each medication. Alternatively, Medicaid insurance may be a marker for other patient characteristics associated with low adherence for which we were not fully able to adjust.

Site differences were also notable in this study; these differences could have been due to differences in insurance coverage in Tennessee versus Massachusetts (which has near‐universal coverage), differences in types of insurance (eg, fewer patients at Brigham and Women's Hospital had Medicaid than at Vanderbilt), cultural and geographic differences between the 2 locations, or other differences in transitional care between the 2 sites.

This study corroborates previous literature on medication adherence (specifically unintentional nonadherence) in the outpatient setting,4, 811 for example, on the association of younger age with low adherence in certain populations. On the other hand, it may contrast with previous literature which has sometimes shown a relationship between patient education or health literacy and medication adherence.14, 15, 2835 However, previous studies have not focused on the transition from inpatient to outpatient settings. Perhaps intensive medication education in the hospital, even under usual care, mitigates the effects of these factors on postdischarge adherence. Finally, baseline adherence seems to correlate with postdischarge adherence, a finding which makes intuitive sense and has been previously reported for specific medications.36

There are several limitations to this study. Although large, the study was performed at only 2 clinical sites where most patients were white and fairly well‐educated, perhaps because patients admitted to a tertiary care center with ACS or ADHF are more affluent than general medical inpatients as a whole; this may limit generalizability. Postdischarge medication adherence might have been higher than in other patient populations given the nature of the population, possible loss‐to‐follow‐up bias, and the fact that half of the subjects received an intervention designed to improve medication management after discharge; such low rates of nonadherence in our study may have reduced our ability to detect important predictors in our models. In addition, the period of follow‐up was 30 days, thus limiting our findings to short‐term postdischarge medication adherence. Postdischarge medication adherence was based on patient self‐report, which not only assumed that the patient was still managing his/her own medications after discharge, but may also be susceptible to both recall and social acceptability bias, which might overestimate our adherence scores, again limiting our ability to detect important predictors of nonadherence. However, other studies have shown a good correlation between self‐reported medication adherence and other more objective measures,37, 38 and recall was only for 7 days, a measure used previously in the literature39, 40 and one designed to reduce recall bias. Systematic underreporting in certain patient populations is less likely but possible.

In the future, research should focus on targeting patients who have low baseline adherence to evaluate the effects of various interventions on postdischarge medication outcomes. Repeating the study in a population with a high prevalence of low health literacy might be illuminating, given that previous studies have shown that patients with low health literacy have less ability to identify their medications and have less refill adherence.29, 30

In conclusion, in patients hospitalized with cardiovascular disease, predictors of lower postdischarge adherence include younger age, Medicaid insurance, and low baseline adherence. It may be prudent to assess baseline adherence and insurance type in hospitalized patients in order to identify those who may benefit from additional assistance to improve medication adherence and medication safety during transitions in care.

In the outpatient setting, medication adherence (defined as percentage of prescribed medication doses taken by a patient during a specific time period) ranges between 40% and 80% for chronic conditions.1 During acute care hospitalization, changes are often made to patients' medication regimens, which can be confusing and contribute to nonadherence, medication errors, and harmful adverse events.2 Indeed, it is estimated that almost half of patients encounter a medication error after discharge, and approximately 12%17% experience an adverse drug event after returning home.36 It is likely that some of these adverse events may be the result of medication nonadherence.7 Improved patientprovider communication, systems to reconcile prehospitalization and posthospitalization medications, as well as development of mechanisms to enhance adherence, may prevent many of these errors and have become new targets for quality improvement.4, 8 Although postdischarge medication adherence is a crucial target for avoiding adverse events and rehospitalization, few studies have focused on understanding its incidence and predictors, in particular, patient demographic factors such as age and insurance status.911

In addition, few studies have looked at general and posthospital adherence in a population where health literacy is measured, an important area because medication changes during hospitalization may be particularly confusing for patients with low health literacy.11, 12 Health literacy is defined as the degree to which an individual has the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions.13 Prior outpatient research shows that low health literacy is associated with poor patient understanding of the medication regimen and instructions for medication use, which may contribute to postdischarge medication nonadherence.14, 15 Understanding the factors associated with postdischarge medication adherence could help refine interventions that are oriented toward improving transitions in care, patient safety, and reducing unnecessary rehospitalization.

We report here on factors associated with postdischarge medication adherence using data from the Pharmacist Intervention for Low Literacy in Cardiovascular Disease (PILL‐CVD) study.16

METHODS

RESULTS

Table 1 shows descriptive baseline patient characteristics of study sample (responders) as well as nonresponders at 30 days. For the responders, the mean age of the 646 patients was 61.2 years, 94.7% were insured, and 19.3% had inadequate or marginal health literacy. Patients were prescribed an average of 8 preadmission medications. Most patients (92.3%) had a regular PCP prior to admission. Nonresponders had nonsignificant trends towards having lower health literacy, medication understanding, and baseline medication adherence.

The average postdischarge adherence score was 95% (standard deviation [SD] = 10.2%), and less than 10% of patients had an adherence score of less than 85%; overall the distribution was left‐skewed. Table 2 illustrates crude and adjusted parameter estimates for variables in the model. Table 3 shows significant findings in the fully adjusted model, which used multiple imputation techniques to account for missing data.

Intervention arm was of borderline statistical significance in predicting postdischarge adherence (P = 0.052), and so was removed from the final model. Study site, age, insurance, and baseline adherence were the only significant independent predictors of postdischarge adherence in the fully adjusted model (Table 3). For example, for every 10‐year increase in age, patients had, on average, an adjusted 1% absolute increase in their adherence score (95% confidence interval [CI] 0.4% to 2.0%). For every 1‐point increase in baseline medication adherence (based on the Morisky scale), there was a 1.6% absolute increase in medication adherence (95% CI 0.8% to 2.4%). In unadjusted analyses, patients with Medicaid were less adherent with medications after discharge than were patients with private insurance. This difference became nonsignificant in adjusted analyses, but when analyses were repeated using multiple imputation techniques, the results again became statistically significantMedicaid insurance was associated with a 4.5% absolute decrease in postdischarge adherence compared with private insurance (95% CI 7.6% to 1.4%). Study site (specifically, Brigham and Women's Hospital) was also a significant predictor of greater postdischarge medication adherence. Years of education was a significant predictor of adherence in unadjusted analyses, but was not an independent predictor when adjusted for other factors. When baseline adherence was removed from the multiple imputation model, there were no changes in which factors were significant predictors of adherence.

DISCUSSION

In this study, we found that low baseline adherence, younger age, Medicaid insurance, and study site were significant predictors of lower 30‐day medication adherence. Of particular interest is our finding regarding baseline adherence, a simple measure to obtain on hospitalized patients. It is notable that in our study, education was not an independent significant predictor of postdischarge adherence, even when baseline adherence was removed from the model. The same is true for medication understanding, cognitive function, and health literacy.

Older patients appeared more adherent with medications in the month after hospital discharge, perhaps reflecting increased interaction with the healthcare system (appointments, number of physician interactions), a greater belief in the importance of chronic medication management, or a higher level of experience with managing medications. A similar relationship between age and adherence has been shown in outpatient studies of patients with hypertension, diabetes, and other chronic diseases.2427

Medicaid patients may be less likely to remain adherent because of the plan's limited coverage of medications relative to patients' ability to pay. For example, Medicaid in Tennessee covers the first 5 generic medications at no cost to the patient but has co‐payments for additional medications and for brand name drugs. Medicaid in Massachusetts has co‐payments of $1 to $3 for each medication. Alternatively, Medicaid insurance may be a marker for other patient characteristics associated with low adherence for which we were not fully able to adjust.

Site differences were also notable in this study; these differences could have been due to differences in insurance coverage in Tennessee versus Massachusetts (which has near‐universal coverage), differences in types of insurance (eg, fewer patients at Brigham and Women's Hospital had Medicaid than at Vanderbilt), cultural and geographic differences between the 2 locations, or other differences in transitional care between the 2 sites.

This study corroborates previous literature on medication adherence (specifically unintentional nonadherence) in the outpatient setting,4, 811 for example, on the association of younger age with low adherence in certain populations. On the other hand, it may contrast with previous literature which has sometimes shown a relationship between patient education or health literacy and medication adherence.14, 15, 2835 However, previous studies have not focused on the transition from inpatient to outpatient settings. Perhaps intensive medication education in the hospital, even under usual care, mitigates the effects of these factors on postdischarge adherence. Finally, baseline adherence seems to correlate with postdischarge adherence, a finding which makes intuitive sense and has been previously reported for specific medications.36

There are several limitations to this study. Although large, the study was performed at only 2 clinical sites where most patients were white and fairly well‐educated, perhaps because patients admitted to a tertiary care center with ACS or ADHF are more affluent than general medical inpatients as a whole; this may limit generalizability. Postdischarge medication adherence might have been higher than in other patient populations given the nature of the population, possible loss‐to‐follow‐up bias, and the fact that half of the subjects received an intervention designed to improve medication management after discharge; such low rates of nonadherence in our study may have reduced our ability to detect important predictors in our models. In addition, the period of follow‐up was 30 days, thus limiting our findings to short‐term postdischarge medication adherence. Postdischarge medication adherence was based on patient self‐report, which not only assumed that the patient was still managing his/her own medications after discharge, but may also be susceptible to both recall and social acceptability bias, which might overestimate our adherence scores, again limiting our ability to detect important predictors of nonadherence. However, other studies have shown a good correlation between self‐reported medication adherence and other more objective measures,37, 38 and recall was only for 7 days, a measure used previously in the literature39, 40 and one designed to reduce recall bias. Systematic underreporting in certain patient populations is less likely but possible.

In the future, research should focus on targeting patients who have low baseline adherence to evaluate the effects of various interventions on postdischarge medication outcomes. Repeating the study in a population with a high prevalence of low health literacy might be illuminating, given that previous studies have shown that patients with low health literacy have less ability to identify their medications and have less refill adherence.29, 30

In conclusion, in patients hospitalized with cardiovascular disease, predictors of lower postdischarge adherence include younger age, Medicaid insurance, and low baseline adherence. It may be prudent to assess baseline adherence and insurance type in hospitalized patients in order to identify those who may benefit from additional assistance to improve medication adherence and medication safety during transitions in care.