Heparin Monitoring and Patient Safety: A College of American Pathologists Q-Probes Study of 3431 Patients at 140 Institutions Open Access

Adequate heparin treatment significantly decreases morbidity and mortality from acute thrombotic disease. Unfortunately, heparin also causes hemorrhagic complications from over-anticoagulation1,2 or heparin-induced thrombocytopenia.3,4 For these reasons, administration of unfractionated heparin is best accompanied by monitoring for anticoagulant and antiplatelet effects, with the dose adjusted in response to monitoring results.

Despite the wealth of literature on heparin administration in controlled research trials, little published information is available about monitoring of heparin therapy in typical practice (“real-world”) settings. Studies published to date generally have been restricted to single academic medical centers, and the study period often has focused on the introduction of a new dosing nomogram, which may have increased awareness of the need for heparin monitoring.

This study examined heparin therapy monitoring at a large cohort of institutions. Inpatients who were treated with unfractionated standard-dose heparin for a minimum of 72 hours were studied to determine whether they received minimal levels of anticoagulant and platelet monitoring and the degree of anticoagulation achieved. The study was also designed to examine whether any of 20 institutional variables related to heparin dosing and monitoring practices were associated with either higher institutional rates of monitoring or more consistent achievement of therapeutic anticoagulation without overshooting into the supratherapeutic range.

The study was conducted according to the Q-Probes study format previously described.5 After pilot testing and refinement of the data collection instrument, College of American Pathologists (CAP) Q-Probes subscribers received mailed data-collection instructions. Participating institutions first answered questions regarding the methods used in their institution to order and monitor anticoagulants. Each institution then reviewed pharmacy records and identified up to 30 inpatients who had been treated with standard-dose, intravenous, unfractionated heparin therapy for at least 3 days (72 hours) during hospitalization. Study patients could be entered into the study on any day of the week, but a maximum of 4 study patients could be entered on any single day. Excluded from this study were outpatients, ambulatory surgery patients, emergency department patients who were not admitted, nursing home patients, and outreach patients. Patients receiving low-dose heparin (5000 units per dose) or low-molecular-weight heparin were also excluded, as were patients discharged before 72 hours with orders to continue receiving heparin as outpatients.

For each study patient, the following data were recorded: (1) date and time of initial heparin dose; (2) whether anticoagulant effect was monitored within 12 hours of the first heparin dose by measurement of activated a partial thromboplastin time (aPTT) or anti–factor Xa; (3) whether a platelet count was monitored within 72 hours of the first heparin dose; (4) whether the patient had at least 1 therapeutic anticoagulant level within the first 24 hours of heparin administration; and (5) whether the patient had 2 or more supratherapeutic anticoagulant levels within the first 72 hours of heparin administration.

To ensure comparability of participant responses across institutions, several definition were established.

For the purposes of this study, aPTT or anti–factor Xa values within a facility's own therapeutic range were considered therapeutic anticoagulation.

For the purposes of this study, an aPTT or anti–factor Xa value greater than the upper therapeutic limit of the facility's therapeutic range.

The shortest time in seconds for aPTT, or the lowest level of anti–factor Xa, that was considered to be an acceptable therapeutic level at the participating institution.

The longest time in seconds for aPTT, or the highest level of anti–factor Xa, that was considered to be an acceptable therapeutic level at the participating institution.

One hundred forty institutions submitted data for this study. Most institutions (98.6%) were in the United States; the remaining 2 were in Canada. Twenty-nine percent of participating institutions were teaching hospitals, and 16% had a pathology residency program. Within the past 3 years, the CAP had inspected 88% of the participants, and 74% had been inspected by the Joint Commission on Accreditation of Healthcare Organizations (JCAHO). Tables 1 and 2 display other characteristics of participating institutions.

Calculations used to derive each of the study variables have been described elsewhere.6 Rates of monitoring and the degree of anticoagulation achieved at each participating institution were tested for association with institutional demographic and practice variable information. P ≤ .05 was considered to be statistically significant.

Four quality measures (dependent variables) were assessed for each institution: (1) the proportion of patients who had at least 1 aPTT or anti–factor Xa level measured within 12 hours of initial heparin administration; (2) the proportion of patients who had at least 1 platelet count performed within 72 hours of the first heparin dose; (3) the proportion of patients who achieved therapeutic anticoagulation within 24 hours of heparin administration, as evidenced by at least 1 aPTT or anti–factor Xa level within the institution's defined therapeutic range; and (4) the proportion of patients who were not over-anticoagulated during the first 72 hours of heparinization, as evidenced by having no or only 1 aPTT or anti–factor Xa level above the institution's defined therapeutic range.

Twenty institutional factors (independent or predictor variables) were identified that could potentially be associated with higher institutional rates of monitoring or anticoagulation control. The 20 factors consisted of the 4 attributes in Table 1, occupied bed size, CAP and JCAHO inspection, the presence of pathology and/or other residency programs, the upper and lower limits of the institution's therapeutic range, and the 10 practice variables listed in Table 2.

The correlation of each of the quality/dependent variables with each institutional predictor variable was assessed separately using nonparametric Wilcoxon rank sum tests and Kruskal-Wallis tests. For each of the quality variables, predictor variables that showed differing rates across their levels at the significance threshold (P < .05) were then introduced into a stepwise multiple regression model to determine which of the institutional factors were independent predictors of the quality variable. Ranked values of the quality variables were used when the original distributions tended to be heavily skewed. All variables remaining in each model were significant at the P < .05 level.

Several participating institutions did not answer all of the questions on the questionnaire. These institutions were excluded only from tabulations and analyses that required the missing data element. A sizable minority of participants (44%–48%, depending on the quality measure in question) could not identify the full complement of 30 heparinized patients for study during the data collection period. Institutions that reported a quality measure for fewer than 10 patients were excluded from the analysis of that measure. This resulted in the exclusion of 13 to 14 laboratories, depending on the measure. An analysis with and without laboratories that reported on 10 to 29 patients did not produce any significant differences in any of the quality measures.

One hundred forty institutions submitted information about 3431 heparinized inpatients. Ninety-five percent of heparinized patients had at least 1 aPTT or anti–factor Xa level obtained within 12 hours of heparin administration; 87% had at least 1 platelet count obtained within 72 hours of heparin administration; 78% had at least 1 therapeutic aPTT or anti–factor Xa level reported within 24 hours of their first heparin dose; and 62% had no or only 1 supratherapeutic aPTT or anti–factor Xa level within 72 hours of heparin administration. Performance varied among the 140 participating institutions, as shown in Table 3. The interinstitutional range in performance for aPTT and platelet monitoring was modest, whereas the range for achieving therapeutic anticoagulation and avoiding supratherapeutic anticoagulation was wider.

Participants were asked to provide the lower and upper limits for the therapeutic anticoagulation range in their laboratory. The data distributions are shown in Table 4; wide interinstitutional variation in therapeutic ranges was noted. The variables in Tables 1, 2, and 4, along with occupied bed size, CAP and JCAHO inspection, and the presence of pathology and other residency programs, were examined to determine which institutional factors, if any, were associated with better performance on the 4 anticoagulation quality measures.

Statistically significant associations are shown in Table 5. No institutional factor was significantly associated with the percentage of patients with at least 1 therapeutic anticoagulation level achieved within 24 hours of the first heparin dose.

We performed a patient-centered analysis to test whether patients who had their first heparin dose on a weekend had worse performance on the 4 quality variables than patients who had their first heparin dose on other days. A χ² goodness-of-fit test was used to evaluate the hypotheses. No significant associations were found.

The coagulation status of patients receiving heparin is in constant flux. At any point in time, the coagulability of a heparinized patient's blood is determined by the amount and timing of heparin the patient has received, the patient's weight and constitutional susceptibility to heparin anticoagulants, and other factors that may also impact the coagulation system, such as an ongoing infection, trauma, concomitant hemostatic defects, the administration of other pharmaceuticals, the presence of disorders that consume coagulants, and idiosyncratic reactions that may develop in response to heparin, such as heparin-induced thrombocytopenia.7,8 The optimal frequency for monitoring heparinized patients varies from patient to patient, and even the most intensive monitoring program provides only snapshots of a patient's dynamically evolving coagulation status.

Despite these limitations, patient safety dictates that certain minimal monitoring standards be applied to all patients receiving standard-dose unfractionated heparin, given the known risk of hemorrhagic complications due to excessive anticoagulation7 and heparin-associated thrombocytopenia.9,10 Similarly, there are accepted therapeutic goals that properly heparinized patients should reasonably meet.7 This study describes how well 140 institutions were able to meet these minimal, safe monitoring standards and achieve reasonable therapeutic goals in 3431 patients.

The 140 institutions' success at monitoring patients was good. At the median hospital, most (98%) patients receiving heparin had their aPTT or anti–factor Xa level tested within 12 hours after therapy was started, and 93% had their platelet counts checked for heparin-induced thrombocytopenia within 72 hours. Even the hospital at the 25th percentile reported that 93% and 82% of its patients had aPTT and platelet testing performed, respectively, within the required time frame.

The 140 institutions had less success at achieving desired levels of anticoagulation in heparinized patients, and there was more variability from site to site in reaching therapeutic objectives. At the median institution, one fifth of heparinized patients had not recorded even 1 therapeutic anticoagulation level during the first 24 hours of heparin administration. The bottom 25th percentile hospital reported that 1 in 3 patients had failed to achieve therapeutic anticoagulation after 1 day of therapy. Even at the top fourth of hospitals, 1 in 10 patients had not achieved therapeutic anticoagulation after 24 hours of therapy.

Of the 4 quality measures we studied, performance was poorest with regard to keeping patients out of the supratherapeutic range. At the median institution, only two thirds of patients stayed out of their institution's supratherapeutic anticoagulation range (as documented by at least 2 supratherapeutic aPTT or anti–factor Xa values in the first 72 hours). At hospitals in the 25th percentile, 2 or more supratherapeutic levels during the first 72 hours of therapy were recorded in 1 of every 2 heparinized patients. Some of these supratherapeutic levels may not be the result of over-anticoagulation, as some study patients may have had supratherapeutic levels due to consumptive coagulopathies that were being treated with heparin. We did not collect information about the indications for heparin administration or underlying disease in study patients, but based on our personal experience we do not believe it likely that many patients in general hospitals who receive unfractionated heparin for at least 72 hours suffer from consumptive coagulopathy.

There was unexpectedly wide variability among sites' definitions of therapeutic and supratherapeutic anticoagulation. Different reagent lots used by study participants might explain some of this variability, but differences between reagents is highly unlikely to account for the wide variation that study participants reported. Ten percent of sites reported their lower aPTT therapeutic limit to be 40 seconds or less, while another 10% reported the lower limit to be 65 seconds or more. Similarly, the upper therapeutic aPTT limit was reported to be 65 seconds or less at 10% of sites, while another 10% reported it to be 116 seconds or more—a difference of almost a minute. This wide variation in reference ranges almost certainly made it easier for some sites to maintain therapeutic anticoagulation than others. Sites that had higher upper limits of their therapeutic aPTT range tended to have fewer patients with 2 or more supratherapeutic values. However, an institution's lower therapeutic limit was not significantly associated with the proportion of patients achieving therapeutic levels within 24 hours.

What can institutions do to improve their performance? We tested a number of demographic and practice variables to determine whether any were associated with better performance. However, none of the comparisons provided much insight into how to improve anticoagulation monitoring, more reliably achieve therapeutic anticoagulation, or keep patients out of the supratherapeutic range. The methods used to order heparin and coagulation tests (computer vs manual order sheets; predefined vs on-the-fly dosing and monitoring schedule) did not influence performance. Weekend performance did not differ from weekday performance. The type of hospital (public vs private) and the presence of a dedicated anticoagulation service were not associated with better performance. Either these variables are not important, or our study lacked the statistical power to detect an association. Patients who were not monitored with aPTT within 12 hours were slightly more likely to not have a platelet count performed within 72 hours, but the association was not particularly strong.

Rural hospitals tended to perform better than nonrural hospitals with respect to aPTT monitoring, but performed worse with respect to platelet monitoring. We do not believe this pair of paradoxical associations provides much insight, even though the associations were statistically significant. Institutions that had not been accredited by JCAHO performed better on aPTT monitoring than those that had been accredited by JCAHO; however, it is difficult to believe that JCAHO inspection could, in and of itself, be a cause of poor performance.

Although data from this study do not suggest specific avenues for improvement, we believe the results of this study can be used to formulate a reasonable performance benchmark. We propose that all institutions seek to perform at least at the level demonstrated by the 75th percentile study participant: 100% aPTT monitoring within 12 hours, 100% platelet monitoring within 3 days, therapeutic anticoagulation achieved within 24 hours in at least 90% of patients, and at least 75% of patients with no or only 1 supratherapeutic level of anticoagulation during the first 72 hours of therapy. In one study of a weight-based heparin-dosing algorithm at a single institution in a research setting, 97% of patients achieved therapeutic anticoagulation within 24 hours.11 Other studies showed rates similar to the 75th percentile hospital in this study.12–14 

The wide interinstitutional variation in therapeutic ranges that we encountered also requires attention. We did not collect information about how institutions established their therapeutic ranges and whether recommended methods were used. For aPTT, calibration against plasma heparin concentrations using protamine titration or an anti–factor Xa chromogenic assay are acceptable methods.7 Unless a laboratory is using reagents with unusual sensitivity or is serving a patient population with unusual needs, the therapeutic aPTT range should be near the median observed in this study, namely, a lower aPTT limit of 50 seconds and upper aPTT limit of 80 seconds. The wide variation in the upper therapeutic limit that we observed in this study leads to the unsettling possibility that patients at some study institutions are at greater risk of bleeding than patients at other institutions.

At least 6 limitations of this study should be born in mind: (1) The quality measures used in this study represent the minimum levels of monitoring that the authors believe appropriate for all patients receiving standard-dose unfractionated heparin during the initial 72 hours of therapy. For many patients, much more stringent monitoring of heparin effect is appropriate. (2) The results of this study are not applicable to patients receiving low-molecular-weight heparin. The more predictable clinical response of low-molecular-weight heparin—which has replaced unfractionated heparin in many institutions—may not require laboratory monitoring at all, although some investigators have advocated monitoring low-molecular-weight heparin with activated factor X, especially in selected populations (eg, children).7 (3) Although there is a large body of evidence linking the clinical effectiveness of heparin to aPTT values, the antithrombotic effect of unfractionated heparin may be more accurately measured with an anti–activated factor X assay.15 Since 135 of the 140 laboratories in this study used aPTT to assess anticoagulant effect, it is theoretically possible that some patients considered outside the therapeutic range in this investigation were, in fact, appropriately anticoagulated. (4) Some of the variation in performance among institutions in this study may have been due to differences in patient mix that we could not measure. For example, poorer performance by some institutions might reflect a greater proportion of heparinized patients with disseminated intravascular coagulation or other unstable conditions that made it more difficult to achieve therapeutic anticoagulation levels. (5) Some of the variability in performance among the participating institutions may have resulted from the wide variation in therapeutic ranges we found among study sties. If some sites did not follow recommended methods to determine their therapeutic range, the proportion of their patients actually out of range may have been overstated or understated. (6) Finally, participation in the CAP Q-Probes program is voluntary, and participating institutions may not be representative of health care facilities in general.

Despite these limitations, this investigation provides what we believe to be the first multi-institutional examination of unfractionated heparin monitoring practices in a real-world setting. The data suggest that most institutions meet minimum standards for monitoring patients receiving unfractionated, standard-dose heparin. However, there is significant interinstitutional variation in how successfully patients are brought into the therapeutic range and kept out of the supratherapeutic zone. There is also significant variation in the therapeutic range used at different institutions, an observation that underscores the importance of using accepted methods to establish therapeutic ranges of both the aPTT and anti–factor Xa assays. Given the important clinical impact of thrombotic disease that can be prevented by adequate anticoagulation, as well as the serious sequela of hemorrhagic complications that result from over-anticoagulation, the variation we observed represents an important opportunity to improve patient safety.