Introduction

The increasing prevalence of type 2 diabetes, and the large health and economic burden of this disease on both patients and society, make type 2 diabetes a pressing public health issue in the United States (US). Quality assessment measures such as those developed by the National Committee for Quality Assurance (NCQA) have been used to evaluate healthcare delivery and management for providers and health plans to guide patient care and address public health issues on a health system level. Even among therapeutic areas classified as “high-impact” based on prevalence, mortality, and costs by the National Pharmaceutical Council, the largest number of direct and indirect NCQA quality measures pertain to diabetes, demonstrating the continued need and national effort to improve diabetes care and management in the US [1].

Hemoglobin A1C (HbA1C) control is an essential part of care for type 2 diabetes and, among other quality measures, is a key component of all quality measure performance assessment. The Healthcare Effectiveness Data and Information Set (HEDIS) defines poor diabetes control as HbA1C >9.0% (75 mmol/mol) and good diabetes control as HbA1C <8.0% (64 mmol/mol) in adults 18–75 years of age. The Physician Quality Reporting System (PQRS) encourages reporting the percentage of diabetes patients with poor control [2].

The value of timely treatment intensification for diabetes patients with above target HbA1C levels has been well-documented. Several studies showed that patients who received more intensive therapy or had their medications adjusted in a timely fashion improved their HbA1C levels [3–6]. The treatment guidelines from the American Diabetes Association (ADA) and American Association of Clinical Endocrinologists (AACE) recommend treatment intensification for patients not at treatment goal after three months of therapy [7, 8]. Even with abundant evidence of the value of treatment adjustment, failure of treatment intensification among patients with above-target HbA1C is commonly observed. One US database study found that median time to treatment intensification was almost two years [9]. An international study with longer follow-up found time to treatment intensification could be greater than seven years [10].

Several factors are associated with diabetes treatment inertia, which is defined as failure to intensify pharmacotherapy in accordance with clinical guidelines. Recent studies using primary care populations suggest that complexity of treatment [11], patient medication adherence [12], HbA1C level [13], specialty of the physician initiating the therapy [14], physician experience level [13], and other provider-, patient- and system-level barriers [15] may be linked to diabetes treatment inertia. While no study has assessed the direct relationship between timely treatment intensification and diabetes quality measure performance, it is reasonable to expect that timely treatment intensification would contribute to greater likelihood of achievement of HbA1C quality performance goals as measured by HbA1C control.

Given the treatment inertia observed in the above studies, increased adherence to treatment guidelines and timely treatment intensification for patients with uncontrolled type 2 diabetes may greatly improve quality measure performance for health plans and providers. Compared to population-level databases, electronic medical record (EMR) data for a registry population from a physician practice would be expected to have more complete data, including labs and vital data, to support an evaluation of this relationship, thereby enhancing both validity and applicability of the results to clinical practice management. In addition, since the EMR is used for daily patient encounters, the data should be of higher quality, although there may still be missing data. This study aimed to assess the relationship between diabetes treatment intensification and HbA1C control quality-of-care measures among patients with uncontrolled type 2 diabetes.

Materials and methods

Study population

This retrospective study used de-identified EMR and diabetes registry data from an ACO located in Bergen County, New Jersey, USA. The patient population served by this practice are fully-insured and primarily are in the middle to upper-middle socioeconomic class. The study period was from January 1, 2011 to June 30, 2016, with a patient index identification period from July 1, 2011 to June 30, 2015. Index date for each included patient was defined as the date of the first above-target HbA1C test (i.e., ≥8.0%) within the index identification period. The baseline period was defined as the six months prior to index and the follow-up period was one year following index (Fig 1). Within the follow-up period, the treatment intensification window was defined as the first 120 days following study index date.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study design and timeframe.

https://doi.org/10.1371/journal.pone.0199011.g001

Included patients were required to a) have physician-diagnosed type 2 diabetes; b) have had an HbA1C level ≥8.0%, which was considered uncontrolled diabetes, during the study period; c) be 18–75 years of age at index date; d) have had at least 6 months of medical records prior to index date (baseline period) and 1 year of medical records after index date (follow-up period); and e) have used at least one oral antidiabetes (OAD) agent during the 6-month baseline period (Fig 2). Patients who were on insulin treatment during the 6-month baseline period were excluded from the study (Fig 2). OAD drug classes examined included biguanides, sulfonylureas, dipeptidyl peptidase-4 (DPP-4) inhibitors, insulin-sensitizing agents, sodium glucose cotransporter-2 (SGLT2) inhibitors, meglitinides, bile acid sequestrants, and alpha glucosidase inhibitors, alone or in combination. As all data were deidentified, institutional review board (IRB) approval was not needed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Selection flow diagram demonstrating identification and derivation of the final patient cohorts.

https://doi.org/10.1371/journal.pone.0199011.g002

Results

A total of 547 patients with type 2 diabetes were identified after applying all inclusion and exclusion criteria; of those, 480 patients had at least one HbA1C test after the treatment intensification window and were included for statistical analyses that incorporated an HbA1C test result (Fig 2). Evaluations of patient baseline characteristics by treatment intensification status (Table 1) demonstrated that a greater proportion of patients in the cohort with poor HbA1C control at the index test received treatment intensification than patients in the cohort with superior HbA1C control (P = 0.0016). BMI was higher among patients who received treatment intensification compared to patients who did not (P = 0.0139). Despite having above-target HbA1c levels at index, only 39.9% of patients received treatment intensification during the 120-day window.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient and provider baseline characteristics.

https://doi.org/10.1371/journal.pone.0199011.t001

After the treatment intensification window, patients’ HbA1C control status was examined by comparing the cohorts with superior and poor control. The odds of receiving treatment intensification were about 1.8 times higher (Table 2) among the cohort with poor HbA1C control in comparison to the cohort with moderate HbA1C control, P = 0.0027. The more types of OADs used at baseline, the less likely patients received treatment intensification, with P values of 0.0058, 0.0102 and 0.0119 for 4, 3 and 2 OAD types, respectively. Patients with higher BMI were more likely to receive treatment intensification (P = 0.0159). No other physician or patient factors were significantly associated with treatment intensification. After accounting for the difference in patient factors, the probability of treatment intensification varied by physician, but the result was not statistically significant. Timely treatment intensification was significantly associated with superior HbA1C control (Table 3) but not with poor HbA1C control (Table 4). Hispanic patients were approximately three-fold more likely than White patients to still have poor HbA1C control after treatment intensification, and male patients were approximately half as likely as female patients to experience poor HbA1C control after treatment intensification (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Physician and patient characteristics associated with type 2 diabetes treatment intensification.

https://doi.org/10.1371/journal.pone.0199011.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association of treatment intensification with superior HbA1C control, using the immediate next level after treatment intensification window (N = 480).

https://doi.org/10.1371/journal.pone.0199011.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Association of treatment intensification with poor HbA1C control, using the immediate next level after treatment intensification window (N = 480).

https://doi.org/10.1371/journal.pone.0199011.t004

Compared with the family medicine physician specialty, internal medicine physician specialty was associated with a lower likelihood of patients having poor HbA1C control after the treatment intensification window (P = 0.0275) (Table 4). A higher index HbA1C level was consistently associated with worse HbA1C control status after the treatment intensification window in all four models. After accounting for differences in patient characteristics, variability between different physician specialties was not significant in any of the four models.

Patients who received treatment intensification had a higher average number of HbA1C tests during their follow-up period compared with patients who did not (mean [SD]: 2.49 [0.89] vs. 2.26 [0.9]; P = 0.0036).

Discussion

After an above-target HbA1C level (≥8.0%), more than 60% of patients with type 2 diabetes still did not receive treatment intensification (Table 1). Patients with higher index HbA1C levels, however, were more likely to receive timely treatment intensification, a result supported by the findings in similar studies [17, 18]. Unlike another study, which showed that younger patients were more likely to undergo treatment intensification, the current study (Table 2) showed that patient age had no bearing on the likelihood of the patient regimen being intensified [17]. In addition, in the current study, patients receiving fewer unique OAD agents were more likely to receive timely treatment intensification (Table 2). These findings fall towards the better end of the range of previously published literature [11]. Previous large retrospective studies suggest that anywhere between 56% and 73% of patients with type 2 diabetes do not receive treatment intensification within 3 to 6 months following the index event [6, 9, 14, 17]. This is similar to the current study, where more than 60% of patients did not receive treatment intensification within 4 months after an above-target HbA1C level ≥8.0%. As shown in the current study and in other previous studies, patients who receive treatment intensification within 6 months have significantly greater HbA1C reduction (-0.33%) within 1 year of follow-up compared to patients who do not receive treatment intensification [6].

A previous study conducted in France by Balkau et al. used electronic records (2008–2009) of a panel of French general practitioners [17]. Similar to the current study, they conducted univariate and multivariate statistical analyses using Cox proportional hazard models to identify associations between patient or practitioner characteristics and treatment intensification. In their study, factors associated with treatment intensification included age 65–79 years (hazard ratio [HR]: 0.71, P<0.001), and age ≥ 80 years (HR: 0.50, P: <0.001), but not age <65 years; treatment with bitherapy (HR: 1.19, P = 0.005) but not tritherapy, all in comparison to monotherapy; the first HbA1c above the recommended target (7.01–8.5 HR: 1.51 [P <0.001]; 8.51–9.5 HR: 2.16 [P <0.001]; ≥9.51 HR: 1.38 [P = 0.04]), and no treatment for hypertension (HR for patients with treatment: 0.82, P = 0.003).

BMI was not examined in the previous study by Balkau et al., but the current study found that BMI, index HbA1C control category and number of OAD types used at baseline were associated with treatment intensification (Table 2). Patients with higher BMI were more likely to receive treatment intensification (P = 0.0159). The odds of receiving treatment intensification were about 1.8 times higher (Table 2) in the cohort with poor HbA1C control than in the cohort with moderate HbA1C control(P = 0.0027). The more types of OADs used at baseline, the less likely it was for patients to receive treatment intensification (Table 2). Our study found that no other physician or patient factors were associated with treatment intensification, similar to the findings of Balkau and colleagues [17].

Differences between the current and the previous study may have been due to country, data used, patient selection criteria and categorization of HbA1C levels. While our study used 8.0%-9.0% to indicate moderate HbA1C control and >9.0% to indicate poor HbA1C control, Balkau et al. utilized categories of ≤ 7.0, 7.01–8.5, 8.51–9.5, ≥9.51.

Most previous studies conducted to identify predictors of treatment intensification used data sources like EMR [17, 19], administrative claims [9], diabetes treatment database [20], pharmacy dispensing data [21] or a combination of survey, EMR and claims data [22]. One study conducted a prospective analysis of patients enrolled in a managed care program [23]. Major differences between the previous studies included baseline HbA1C levels of the included patients, OAD or insulin use at baseline [9, 17], inclusion/exclusion criteria, the gap between HbA1C tests [17], choice of variables considered and type of statistical analyses conducted. These studies revealed that treatment intensification was associated with the number of OAD types used at baseline [17, 23], younger age [17, 19, 22], higher HbA1C values at baseline [9, 17, 19, 22], longer duration of diabetes [19, 20, 22], point-of-service insurance, mental illness, endocrinologist visit at baseline [9], no treatment for hypertension [17], using insulin at baseline [19], non-adherence to OADs [21], worse glycemia, and patients with longer intervals between routine hyperglycemic visits [17, 23]. Most studies indicated no association between physician-related factors and treatment intensification [17, 19]. Studies also concluded that higher baseline HbA1C, older age, black ethnicity, lower income, and more physician visits were associated with improvement in HbA1C levels after treatment intensification [22]. As indicated in other studies, the current study also showed that both ethnicity (Hispanic) and being female were significantly associated with poor HbA1c control after treatment intensification (Table 4).

In addition, after controlling for available patient and physician factors, no significant variability on the physician level was found in our study, except that patients followed by internists were about half as likely as those followed by family medicine specialists to exhibit poor HbA1c control (Table 4). This is similar to other published literature, which showed no difference in HbA1C control between patients treated by family medicine specialists compared to endocrinologists, [24]; this finding may be attributed to the fact that the physicians examined in the study were from the same practice facility. Furthermore, the fact that this practice attempts to follow the recommended treatment guidelines may have reduced the variability in process and outcomes of patient care. Nonetheless, the study did confirm that timely treatment intensification was significantly associated with superior patient HbA1C control during follow-up. Although an association was not observed for the cohort with poor HbA1C control, this is very likely due to the fact that the study used “patient had an HbA1C level ≥8.0%” as the criterion to trigger treatment intensification instead of using >9.0% levels.

In response to the current study’s findings, the practice has instituted a program using the health maintenance section and practice analytics module of their EMR. They have programmed the system to flag patients with elevated HbA1C levels and send out reminder letters. They also have a “care coordinator” directly contact those patients with elevated HbA1C levels who have not been seen or do not have a scheduled appointment within 3 months to schedule follow-up.

There were several limitations to the study, beginning with the data being from a regional ACO. Although the patient population was diverse in its demographics, care must be taken in generalizing the results. In addition, the study used HbA1C levels ≥8.0% as the selection criterion for the base patient population. This is consistent with the practice’s internal recommendations for treating patients with type 2 diabetes and the outcome measure of HbA1C control status; however, it may not necessarily be closely related to the HbA1C poor control performance measure. ADA and AACE guidelines both recommend treatment intensification for patients not at treatment goal after 3 months of therapy [7, 8]. Considering the treatment patterns in real-world practice, the study used a 120-day treatment intensification window rather than a 90-day window. In addition, it was possible that patients may not have filled the prescription prescribed as part of the treatment intensification regimen and, in that case, treatment inertia could be underestimated. Unfortunately, the EMR data did not include patient fill, thereby preventing determination of treatment adherence. At the same time, treatment intensification that was missing from the EMR data could be categorized as treatment inertia and cause overestimation of treatment inertia. Even so, the data did provide a very valuable perspective that enabled this evaluation of clinicians’ prescribing practices and HbA1C control. Regardless of fill status, these results support an association between the two and add credibility to the value of outcome-based quality measures. Lastly, it would be very hard to experimentally determine if clinical contact causatively results in better HbA1C outcomes, as only motivated patients may consent. The results here suggest the value of regular clinical contact and monitoring, which was associated with better HbA1C control.

This research is the first published instance of primary care EMR data used to evaluate treatment intensification and its effect on diabetes control and adherence to quality measures, specifically HbA1C. The study shows that it is possible to collect and analyze these data to determine how to positively influence physician behavior to improve patient outcomes. Unlike previous studies, one of the key advantages of the current study is the fact that the data set was very complete. The Chief Medical Officer of the practice has been a proponent of EMR use for the past 20 years and all prescriptions that go out must go through the EMR. Given that this retrospective study evaluated a relatively affluent suburban population and that there were 7 endocrinologists on staff at the practice during the study period, it is daunting that the treatment inertia was as high as it was—underscoring the intractability of timely treatment for patients with type 2 diabetes. There are likely psychosocial and other unmeasured patient and physician factors at play here, which should be more closely examined in future studies.

Although researchers typically rely on large datasets available commercially, the current research showed results consistent with previous publications using a dataset available to practicing clinicians on a daily basis in their practices. The use of a practice’s patient data for important research such as this demonstrates that these types of analyses are not restricted to researchers with large data sets, but can also be used to evaluate outcomes for individual clinicians and practices.

Conclusions

Timely treatment intensification was significantly associated with greater likelihood of patients achieving superior HbA1C control (<8.0%) and better HbA1C control quality performance for the practice. Even in a large group practice with resources dedicated to diabetes control, it is incumbent upon clinicians to readily identify and open dialogues with patients who may benefit from closely supervised, individualized attention.

Supporting information

Acknowledgments

At the time of this research, two of the authors (JJS and SF) were employed by AstraZeneca. Quorum Consulting, Inc., now Navigant Consulting, Inc., where two of the authors were employed (RGA and SY), received a grant to conduct this research. All authors contributed significantly to the manuscript in terms of concept design, study design and manuscript development. This does not alter our adherence to PLOS ONE policies on sharing data and materials.