This study was developed and approved by the Steering Committee of the Translating Research in Action for Diabetes (TRIAD) Study and conducted in 1 of TRIAD’s 6 Translational Research Centers, Kaiser Permanente Northern California (KP). KP is an integrated health care delivery system providing comprehensive medical care to a diverse population of approximately 3.2 million members in Northern California.43
Patients were selected for the study from the KP diabetes registry if they were identified as having diabetes before January 1, 2005 and were continuously enrolled with an active drug benefit during all of 2004 and 2005. Continuous enrollment was required so that adherence and treatment intensification could be accurately assessed throughout 2005. Eligible patients were further assessed for the presence of clinically recognized hypertension and hyperlipidemia before January 1, 2005 (see the Appendix
for definitions) using KP automated clinical databases. The validity and reliability of the KP diabetes registry; its laboratory, clinical, and pharmacy databases; and the utility of these databases for assessing both treatment intensification and medication adherence have been documented.44,45
Definitions of Target Levels Patients were defined as being above target levels of A1c if they had an A1c laboratory value ≥7.0% at any point during 2005, using the first measurement in 2005 for each patient if multiple above target values were noted. Similarly, those diabetes patients with hyperlipidemia were defined as above target for LDL-c if they had any LDL-c value ≥100 mg/dL during the year. Those with hypertension were defined as above target for blood pressure if they had at least 2 consecutive SBP readings of ≥130 mm Hg. Two consecutive readings were required because of the greater lability of blood pressure measures; 1 measure was considered sufficient for LDL-c and A1c as HEDIS guidelines recommend 1 LDL-c and A1c test per year. To assess if the relative importance of adherence versus clinical inertia observed at these risk factor target cut points was similar at less stringent cut points, in a second sensitivity analysis, we defined being above target as A1c ≥8.0%; LDL-c ≥130 mg/dL; and 2 SBP readings of ≥140 mm Hg.
Adherence to Medications
Adherence to medications was calculated with KP prescription databases using the validated continuous, multiple interval measure of gaps in therapy (CMG) method.46–48
Validation research has shown the CMG method to be significantly associated with objective measures of medication use such as serum/urine drug levels, physiological drug effects such as blood pressure, and increased comorbidity and cost.47
This method is defined as the proportion of days the patient should
have been on medication therapy during which the patient did not have medication available. For each individual condition (hyperglycemia, hypertension, and hyperlipidemia), CMG was first calculated separately for each medication class filled at least twice in the 12 months before the last date above target levels were observed in 2005. Individual class adherence was then combined into a single measure for all medications prescribed for a single condition, weighting the estimate for each medication class by the number of days from the first to last fill in the 12-month period. Medications filled only once were not included in the analysis because CMG cannot be calculated from single fills. Because many prior studies have found significant clinical effects when cumulative days of refill gaps equal or exceed 20%,42,49,50
we defined poor adherence for each condition as a weighted nonadherence measure of ≥20% across all medications prescribed for the condition.
Treatment intensification was assessed for each condition from KP prescription databases during the 3 months before and the 3 months after first measurement of above target levels in 2005. Intensification was defined as any one of the following 3 occurrences: (a) an increase in the number of drug classes; (b) an increase in the daily dosage of at least 1 ongoing drug class; or (c) a switch to a medication in a different drug class. Seven medication classes were tabulated for hypertension (ace inhibitors, angiotensin antagonists, beta adrenergic blockers, calcium channel blockers, thiazides/related diuretics, potassium-sparing diuretics, and loop diuretics); 5 for hyperlipidemia (statins, bile acid resins, fibrates, niacin, and ezetimibe), and 4 for diabetes (sulfonylureas, metformin, thiazolidinediones, and insulin);44
combination pills were considered as consisting of 2 classes. Daily dosages were categorized as low (near initial starting doses), medium (maintenance range), or high (above maintenance range) based on package insert recommendations and inspection of actual dosage distributions. This analysis focuses on treatment intensification in patients with no evidence of poor adherence, as these patients are the most appropriate targets for treatment intensification interventions and prescribing additional medications for patients in poor adherence may be less effective.
To allow a full 3 months follow-up, we extended the observation into 2006 for those whose above target values were first identified in the final quarter of 2005, excluding the small number of these patients who were no longer with KP in 2006. In assessing both medication adherence and treatment intensification in diabetes, we excluded patients who were using insulin at the time above target A1c levels were noted because neither can be accurately identified in prescription databases.
Data Analyses For each condition separately (hyperglycemia, hyperlipidemia, and hypertension), we first determined the proportions of patients above versus at or below target levels. Among patients who were above target, we then determined the proportions with poor adherence to medications, no current medications or evidence of poor adherence but no treatment intensification, or no evidence of poor adherence and treatment intensification within 3 months before to 3 months after the observation of above target risk factor levels. As adherence and treatment intensification for insulin are both difficult to assess using automated databases, we limited our analysis of hyperglycemia to patients who were not on insulin at baseline. In measuring treatment intensification, we also examined the proportion who were either already on, or whose treatment intensification moved them to, maximal medical therapy (MMT). MMT was defined as: 4 classes of antihypertensive drugs for hypertension; ezetimibe (at any dose), 80 mg simvastatin, or 40 mg atorvastatin for hyperlipidemia; and any insulin starts for diabetes (as exact insulin dosages could not be assessed).
To assess whether those patients who had no evidence of poor adherence and received a timely treatment intensification were more likely to achieve target risk factor values within 6 months than either patients in poor adherence or had not received a treatment intensification, we looked ahead 6 months after the determination of above target risk factor levels for each patient to see what proportions achieved targets according to the definitions given above.
All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC, USA). This study was reviewed and approved by Kaiser Permanente’s Institutional Review Board.