Jeffrey A. Linder, MD, MPH

To compare antibiotic prescribing among retail clinics, primary care practices, and emergency departments (EDs) for acute respiratory infections (ARIs): antibiotics-may-be-appropriate ARIs (eg, sinusitis) and antibiotics-never-appropriate ARIs (eg, acute bronchitis).

 We analyzed retail clinic data from the electronic health records of the 3 largest retail clinic chains in the United States, and data on visits to primary care practices and EDs from the nationally representative National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey.

 Using multivariate models, we estimated an adjusted antibiotic prescribing rate for each site of care, controlling for differences in patient characteristics and diagnosis.

From 2007 to 2009 in the United States, there were 3 million, 167 million, and 29 million ARI visits at retail clinics, primary care practices, and EDs, respectively. For all ARI visits, the adjusted antibiotic prescribing rate at retail clinics (58%) was similar to the rate at primary care practices (62%; 

= .48). For antibiotics-may-be-appropriate ARI visits, the adjusted antibiotic prescribing rate (95%) at retail clinics was higher than at primary care practices (85%; 

<.01). For antibiotics-never-appropriate ARI visits, the adjusted antibiotic prescribing rate (34%) at retail clinics was lower than at primary care practices (51%; 

Compared with primary care practices and EDs, there was no difference at retail clinics in overall ARI antibiotic prescribing. At retail clinics, antibiotic prescribing was more diagnosis-appropriate.

There is concern that retail clinics could increase inappropriate antibiotic prescribing because they have financial incentives to overprescribe medications and they employ nurse practitioners. In the first evaluation of this question using medical record data, we found that the overall antibiotic prescribing rate was similar among retail clinics, physician offices, and emergency departments; furthermore, care at retail clinics was more guideline-concordant. For example, in the case of conditions for which antibiotics are never indicated, retail clinics had a lower antibiotic prescribing rate.

		There are concerns that retail clinics provide an inferior quality of care, but our results do not support these concerns.

		No difference was found in quality of antibiotic prescribing among retail clinics, physician offices, and emergency departments.

Acute respiratory infections (ARIs) such as pharyngitis, rhinosinusitis, and bronchitis are the most common symptomatic reasons why patients seek care in the United States.

ARIs account for half of all adult antibiotic prescriptions and three-fourths of all pediatric antibiotic prescriptions.

Physicians overprescribe antibiotics for ARIs and unnecessarily use broad-spectrum antibiotics, both of which lead to increased antibiotic resistance, adverse drug reactions, and increased cost.

A growing fraction of ARI visits occur at retail clinics, which are located in drugstores and grocery stores and are typically staffed by nurse practitioners (NPs) as opposed to physicians. Retail clinics provide walk-in care for a limited set of conditions for which they have treatment protocols, including ARIs, other minor illnesses, immunizations, sports physicals, and other screening services—ARIs account for 60% of all retail clinic visits.

 Providers at retail clinics independently manage and prescribe medications. State regulations vary on whether NPs at retail clinics must maintain a collaborative or supervisory relationship with a physician.

Currently, there are almost 6 million yearly visits to the 1600 retail clinics in the United States,

and the number of retail clinics is expected to grow rapidly in the coming years.

The impact of this shift to retail clinics in the care of ARIs, and subsequent antibiotic prescribing, is unknown. It is possible this shift may worsen existing problems with ARI antibiotic prescribing patterns, as most retail clinics are owned by for-profit pharmacy chains that have a financial incentive to provide prescription medicines.

Alternatively, it is possible this shift will improve antibiotic prescribing, as NPs often provide care that is more consistent with guidelines

 and retail clinics strictly incorporate evidence-based guidelines into their electronic health records (EHRs).

In prior studies comparing rates of antibiotic prescribing, little difference was observed between retail clinics and outpatient physician practices

; however, these were small studies that used a limited set of diagnoses and health plan claims data. In this study, we compared antibiotic prescribing at retail clinics, primary care practices, and emergency departments (EDs) for all ARIs using national data collected directly from the EHR or the medical chart.

We obtained de-identified data from all visits from 2007 to 2009 (2007, 0.8 million visits; 2008, 2.4 million; 2009, 5.4 million) from 3 retail clinic operators: MinuteClinic, TakeCare, and LittleClinic. Together they operate 81% of all US retail clinics.

These data came directly from the operators’ EHRs.

We obtained 2007 to 2009 data (to match the years of available retail clinic data) on visits to ambulatory practices from the National Ambulatory Medical Care Survey (NAMCS) and outpatient hospital practices from the National Hospital Ambulatory Medical Care Survey (NHAMCS). NAMCS and NHAMCS employ a complex sampling methodology to obtain a nationally representative estimate of outpatient visits in the United States, and the 2 surveys have been used in many studies to characterize physician care patterns in the United States.

 However, neither survey includes retail clinics. For each visit, standard 1-page encounter forms are completed by the clinician, office staff, or an outside coder using the medical chart.

We defined primary care visits in NAMCS as those associated with the physician specialties of family/general practice, internal medicine, or pediatrics, and in NHAMCS as visits to general medicine and pediatric clinics. We excluded the small number of visits with a nurse practitioner because we did not want to confound training and practice setting; we included these visits in a sensitivity analysis, and it had no impact on our findings. To exclude follow-up visits for ARIs, we limited our sample to visits where the treating provider indicated that the major reason for the visit was a new problem. After these exclusions, we sampled 6289 ARI visits. We also conducted a sensitivity analysis in which we examined all ARI visits, including those where it was not a new problem, and we found no substantive differences.

We obtained data on 7629 sampled ED visits for ARIs from 2007 to 2009 from the ED component of the NHAMCS survey. Similar to the physician outpatient visits in NAMCS, a standard encounter form is completed for each visit using the medical chart.

Across the 3 data sources, we looked at patient demographics, reasons for visit, diagnoses, and medications prescribed. We did not include patient temperature because for almost half of ARI visits at primary care practices, the temperature field was blank. Patient drug allergies were not available.

International Classification of Diseases, Ninth Revision, Clinical Modification

we divided the 6 ARI diagnoses into “antibiotics-may-be-appropriate” diagnoses (streptococcal pharyngitis [034.x], otitis media [381.x, 382.x], sinusitis [461.x]) and “antibiotics-never-appropriate” diagnoses (upper respiratory infection [460.x, 465.x], nonstreptococcal pharyngitis and laryngitis [462.x, 464.x], bronchitis [466.x, 490.x, 491.21]). These definitions of antibiotic-appropriateness are forgiving to prescribing of antibiotics. For example, although we consider sinusitis an antibiotics-may-be-appropriate diagnosis, it is not clear that sinusitis ever requires antibiotic treatment.

Excluding Visits With Competing Diagnoses

Patients may have multiple problems that require antibiotics, and the fraction of ARI patients with such competing diagnoses might differ across the 3 care sites. To address this potential concern, we excluded visits with a diagnosis code for a competing diagnosis that might require antibiotics (list of 

). This excluded 3%, 6%, and 10% of the ARI visits at retail clinics, physician offices, and EDs, respectively. In a sensitivity analysis, including such visits had no substantive impact on the results.

Among all eligible ARI visits, we identified any oral antibiotic prescription associated with the visit. In the retail clinic, data prescriptions are identified using the National Drug Code system, while in the NAMCS and NHAMCS surveys, prescriptions are classified using the Multum Lexicon system. We defined broad-spectrum antibiotics as azithromycin and clarithromycin, quinolones, amoxicillin- clavulanate, ampicillin-sulbactam, and second- and third-generation cephalosporins.

We calculated the antibiotic prescribing rate for all ARI visits, for antibiotics-may-be-appropriate diagnoses, for antibiotics-never-appropriate diagnoses, and for each individual ARI diagnosis. We limited our analysis to patients 2 years and older because retail clinics rarely see children under that age.

We used multivariable models to assess the independent association between site of care and antibiotic prescribing, where the outcome variable is a binary (yes/no) indicator for whether an antibiotic was prescribed. Building on prior work,

we combined the data from the 3 clinical sites into a single data set. In the physician office and ED data, each visit is given a sampling weight so that one can calculate nationally representative estimates. Since retail clinics represent a full national sample of visits, in the analyses, these visits were given a sampling weight of 1. The NAMCS and NHAMCS samples include identifiers allowing for control for clustering within individual clinicians and practices and EDs. For the retail clinic data, we could not identify individual clinicians and practice sites associated with a visit; we therefore accounted for a higher level of clustering within each of 3 retail clinic chains. Because it assumes that all retail clinic visits within a given retail clinic chain are similar, the intercluster correlation was large (0.02) and therefore standard errors were conservatively large.

We used separate models for all ARI visits, antibiotics-may-be-appropriate ARI diagnosis visits, and antibiotics-never-appropriate ARI diagnosis visits. Among all ARI visits where an antibiotic was prescribed, we created an analogous logistic regression model where the dependent variable was broad-spectrum antibiotic prescription. In each of the multivariate models we adjusted for gender, age, site of care, and whether the patient had a chronic illness. In the model where we focused on all ARI visits, we also adjusted for diagnosis type (antibiotics-may-be-appropriate diagnoses vs antibiotics-never-appropriate). We conducted sensitivity analyses where we: 1) did not adjust for diagnosis; 2) adjusted for the 6 specific ARI diagnoses instead of diagnosis type to provide a higher level of specificity; 3) examined only children (aged ≤17 years) as this is an important subpopulation; 4) examined only adults; and 5) examined only patients without a reported chronic illness, as this is the population most likely to be seen at retail clinics. The odds ratio results from all multivariate models are reported in the eAppendix.

To facilitate interpretation of the regression results, we used the method of predictive margins (also called recycled predictions) to report the predicted antibiotic prescribing rate and the predicted broad-spectrum antibiotic prescribing rate for each of the 3 care sites while adjusting for other covariates.

 Specifically, we used the estimated coefficients from the multivariate regression models to predict antibiotic prescribing for each patient visit, alternatively assigning all visits to a given value of the variable (eg, retail clinic, primary care practice, ED), but leaving all other explanatory variables at their original values. We then averaged the predictions across all patient visits. This allowed us to estimate the mean antibiotic prescribing rate if all patient visits occurred at a given site, but otherwise retained the original values of all their other characteristics. 

values less than .05 were considered significant. All analyses were performed using SAS 9.2 (Cary, North Carolina).

Between 2007 and 2009, there were 3 million, 162 million, and 29 million ARI visits, respectively, at retail clinics, primary care practices, and EDs. There were several notable differences in patient characteristics across the 3 care sites (

). Children and adolescents (aged ≤17 years) made up a smaller fraction of ARI visits to retail clinics (27%) than to primary care practices (50%; 

<.01). A smaller fraction of patients reported a chronic illness at retail clinics (4%) compared with primary care practices (20%; 

The mix of ARI visits varied by site of care. At retail clinics, a higher proportion of ARI visits (53%) led to an antibiotics-may-be-appropriate diagnosis (sinusitis, otitis media, and streptococcal pharyngitis) than at primary care physician offices (32%; 

Unadjusted Differences in ARI Antibiotic Prescribing

Not adjusting for other factors, the fraction of all ARI visits at which an antibiotic was prescribed was similar at retail clinics (66%), primary care practices (62%; 

= .28 comparison with retail clinics), and EDs (60%; 

= .11) (Table 2). This pattern varied by antibiotic appropriateness. For antibiotics-may-be-appropriate diagnoses, retail clinic providers were more likely to prescribe antibiotics (95%) than primary care physicians (85%; 

<.01). In contrast, for antibiotics-never-appropriate diagnoses, retail clinics providers were less likely to prescribe antibiotics (33%) than primary care physicians (50%; 

<.01). One notable exception among antibiotics-never-appropriate diagnoses was for acute bronchitis/bronchiolitis where antibiotic prescribing at retail clinics, physicians, and EDs was 77%, 76%, and 70%, respectively. Among all ARI visits where an antibiotic was prescribed, retail clinic providers were less likely to prescribe a broad-spectrum antibiotic than primary care practices (40% vs 49%; 

<.01), but there was no difference between retail clinics and EDs (40% vs 44%; 

Adjusted Differences in ARI Antibiotic Prescribing

The adjusted antibiotic prescribing rate for all ARI visits at retail clinics (58%) was similar to the adjusted antibiotic prescribing rate at primary care practices (62%; 

). For antibiotics-may-be-appropriate ARI visits, the adjusted antibiotic prescribing rate for retail clinics (95%) was higher than at primary care practices (86%; 

<.01). For antibiotics-never-appropriate ARI visits, the adjusted antibiotic prescribing rate at retail clinics (34%) was lower than at primary care practices (50%; 

 <.01). For ARI visits at which an antibiotic was prescribed, the adjusted rate of broad-spectrum antibiotic prescribing at retail clinics was 42%, lower than the rate at primary care practices (49%; 

= .04) but not different from the rate at EDs (44%; 

The differences in antibiotic prescribing were consistent in both the children-only and adult-only analyses (eAppendix Table 2). If we do not account for ARI diagnosis in the model then there is no statistically significant difference in overall ARI antibiotic prescribing between the 3 care sites (eAppendix Table 2). When we only look at patients without a chronic illness, the differences in antibiotic prescribing were similar to those in the model with all visits (eAppendix Table 2).

In a large, nationally representative sample of ARI visits, retail clinics had an antibiotic prescribing rate similar to that of primary care practices and EDs. This is reassuring evidence that the shift in care to retail clinics will not negatively impact antibiotic prescribing in the United States. The lack of a difference is consistent with the larger literature, that demonstrates that nurse practitioners and physicians provide care of similar quality.

While there is no difference across the sites in antibiotic prescribing rate for all ARI visits, there is a difference in both the mix of ARIs seen and in antibiotic prescribing by type of ARI. At retail clinics, half of ARI visits are for antibiotics-may-be-appropriate diagnoses (sinusitis, otitis media, and streptococcal pharyngitis). In contrast, onethird of ARI visits at physician offices and EDs are for antibiotics-may-be-appropriate diagnoses. For antibiotics-never-appropriate diagnoses, primary care practices have a much higher antibiotic prescribing rate. However, for antibiotics-may-be-appropriate diagnoses, we see the opposite pattern—retail clinics have a much higher antibiotic prescribing rate.

These results could be driven by how providers choose a diagnosis. Ideally, a provider makes a diagnosis that leads to the decision to prescribe antibiotics. However, it is also possible that providers first decide whether to give antibiotics and later code a diagnosis based on the decision to prescribe antibiotics. If the latter pattern is common at retail clinics, then it might imply there are no differences in true ARI mix across the sites but that retail clinic providers are more likely than primary care and ED physicians to list an ARI diagnosis for which antibiotics are appropriate (eg, listing sinusitis as the diagnosis instead of a viral upper respiratory condition). In such cases, retail clinic providers are not truly providing care that is more concordant with the diagnosis, as the diagnosis itself might be incorrect.

It is also possible that the difference in ARI mix across the care sites reflects true differences in patient populations. For example, it could be that patients with prior episodes of sinusitis self-select to a retail clinic, and this is why we see a higher fraction of antibiotics-may-be-appropriate ARI diagnoses at retail clinics. If this is the case, then the more diagnosis-concordant pattern we observe at retail clinics means that antibiotics are being reserved for those whom antibiotics might benefit. At many retail clinics, evidence-based guidelines are incorporated into the EHR; if providers choose to go outside the guidelines, the provider must provide a justification.

 This forcing mechanism may lead to decreased antibiotic prescribing and thus higher quality care.

There were differences in the type of antibiotics prescribed at retail clinics, primary care practices, and EDs. Across all ARI diagnoses, controlling for other factors, primary care practices were more likely than retail clinics to use broad-spectrum antibiotics such as azithromycin and quinolones. This pattern is concerning given that these broad-spectrum antibiotics often do not have higher efficacy, and their overuse could increase the prevalence of antibiotic-resistant bacteria. For example, the fraction of pneumococcal serotypes resistant to macrolides is 35.3%

Articles

There are concerns that retail clinics provide inferior quality of care. The authors found no difference in quality among retail clinics, physician offices, and emergency departments.

Antibiotic Prescribing for Respiratory Infections at Retail Clinics, Physician Practices, and Emergency Departments

To determine whether clinical decision support (CDS) is associated with improved quality indicators and whether disabling CDS negatively affects these.

Using the 2006-2009 National Ambulatory and National Hospital Ambulatory Medical Care Surveys, we performed logistic regression to analyze adult primary care visits for the association between the use of CDS (problem lists, preventive care reminders, lab results, lab range notifications, and drug-drug interaction warnings) and quality measures (blood pressure control, cancer screening, health education, influenza vaccination, and visits related to adverse drug events).

There were an estimated 900 million outpatient primary care visits to clinics with EHRs from 2006-2009; 97% involved CDS, 77% were missing at least 1 CDS, and 15% had at least 1 CDS disabled. The presence of CDS was associated with improved blood pressure control (86% vs 82%; OR 1.3; 95% CI, 1.1-1.5) and more visits not related to adverse drug events (99.9% vs 99.8%; OR 3.0; 95% CI, 1.3-7.3); these associations were also present when comparing practices with CDS against practices that had disabled CDS. Electronic problem lists were associated with increased odds of having a visit with controlled blood pressure (86% vs 80%; OR 1.4; 95% CI, 1.3-1.6). Lab result notification was associated with increased odds of ordering cancer screening (15% vs 10%; OR 1.5; 95% CI, 1.03-2.2).

The use of CDS was associated with improvement in some quality indicators. Not having at least 1 CDS was common; disabling CDS was infrequent. This suggests that meaningful use standards may improve national quality indicators and health outcomes, once fully implemented.

		The use of some electronic health record clinical decision support (CDS) was associated with improved quality of care among national primary care outpatient visits from 2006 to 2009.

		Most practices were missing at least 1 CDS module considered essential by meaningful use standards.

		Meaningful use standards may have a significant impact on national quality of care and health outcomes once fully implemented.

The CMS Electronic Health Record Incentive Program, better known as “meaningful use,” provides initial monetary incentives for adoption of electronic health records (EHRs) and subsequently imposes monetary penalties if physician practices do not meet requirements. The meaningful use program has spurred a significant rise in the number of office-based physicians reporting EHR use, from 24% in 2005 to 48% in 2009 to 72% in 2012.

The meaningful use goals and objectives are divided into 3 stages. Stage 1 requirements include using a certified EHR to capture and share patient data by 2015; stage 2, beginning in 2014, includes use of advanced care processes with clinical decision support (CDS); and stage 3, to begin in 2016, will require providers to demonstrate outcome improvements. The time frame for stages 2 and 3 was recently extended.

 As of March 2013, 73% of eligible providers had registered to participate in the meaningful use incentive program and 44% of eligible providers had received incentive payment for meeting meaningful use objectives.

Although the meaningful use requirements have already been established, the evidence is inconsistent regarding improvement in healthcare processes or patient outcomes as a result of the implementation of general and individual EHR components. Prior studies have shown that EHR-based CDS is associated with improved prescribing safety,

 CDS has also been associated with some improvements in quality indicators,

 Despite having EHRs, many physicians report being unable to complete basic panel management activities,

which affects their ability to deliver high-quality care for patients with chronic conditions.

Notwithstanding the mandate to use certified EHRs, use of an EHR in the outpatient setting has not yet been shown unequivocally to improve quality indicators across a broad range of practices and EHRs, especially because there is substantial variability among the 1600 certified EHR platforms.

In addition to the large number of EHRs, they also can be customized to each practice’s needs and resource availability, so results will vary between practices in terms of impact on care even with the use of a single vendor’s practice. Certain types of practices, such as smaller community-based practices, may be less able to fully implement a complete EHR with clinical decision support, which is the kind postulated to have the greatest potential to improve quality.

 Decisions about choices around the CDS functions implemented may affect the quality of care delivered by that practice. To further inform the potential effect of meaningful use stages 2 and 3, we examined the association between having CDS, not having CDS, and disabling CDS, and quality of care indicators using nationally representative data just prior to the start of the meaningful use incentive program.

We performed a retrospective, cross-sectional analysis of adult primary care ambulatory clinic visits in the National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) outpatient department records from 2006-2009 in order to analyze data with little impact from current meaningful use incentive payments. We examined a number of patient and practice characteristics to determine any differences in access to practices with EHRs based on prior work showing associations between these characteristics and EHR availability.

On the premise that most preventive care takes place through a primary care practice, we limited our sample to just those practices.

The NAMCS and NHAMCS are conducted by the Ambulatory Care Statistics Branch of the National Center for Health Statistics (NCHS) of the CDC. These surveys focus on outpatient visits to non—federally funded US medical practices.

The NCHS uses a complex multistage probability sampling design based first on geography, then physician specialty, and finally patient visits to individual practices. Information is collected at ambulatory visits by participating practices or US Census Bureau representatives. The NCHS assigned weights to each visit record to allow for estimation of national statistics. The NCHS institutional review board approves the protocols for the NAMCS and NHAMCS, including a waiver of informed consent for patients.

The NAMCS collects information at a visit level from community-based physician practices; from 2006 to 2009, approximately 30,000 community-based office visits were recorded annually. The NHAMCS collects information at a visit level from hospital-based physician practices; from 2006 to 2009, approximately 34,000 hospital-based office visits were recorded annually.

In all years included in the analysis, both the NAMCS and NHAMCS included questions in their intake surveys that asked whether the practice had an electronic medical record (for other than billing purposes), a computerized patient problem list, computerized warnings of drug interactions or contraindications, a computerized system to view lab results, a computerized system that highlights out-of-range lab values, and a computerized system for reminders for guideline-based interventions and/or screening tests. Response variables indicated that a practice had a function (“yes”), did not have a function (“no”), had the function but disabled it (“turned off”), or was unsure if the practice had the function (“don’t know” or “blank”).

The NAMCS and NHAMCS record health information for each visit sampled. The respondent indicates the top 3 reasons for the visit, which are then coded by NCHS; the patient’s vital signs, including blood pressure; whether any health education was ordered or provided during the visit; and whether an influenza vaccination was ordered or administered during the visit. The respondent also indicates any screening services ordered or performed during the visit. The respondent can record mammography, Pap smears, and scope procedures (NCHS records and codifiers, free text scope procedures, 1 of which is sigmoidoscopy/ colonoscopy), among others.

The NAMCS and NHAMCS collect patient and practice demographic information: patient age, patient race, patient income category (imputed based on patient zip code by NCHS), primary reason for the visit, region of the country, practice ownership, and number of physicians in the practice (solo or non-solo). Race data are collected in the manner already in use by the practice. Due to high nonresponse rates for race (24% to 33% in NAMCS, and 12% to 15% in NHAMCS for the years included), the NCHS also provides a race variable imputed based first on the patient’s location, then on physician specialty and 

 code for primary diagnosis, and finally on a random basis.

Our predictors of interest were EHR CDS modules mandated by stage 1 of meaningful use: electronic problem lists, preventive care reminders, lab result reporting, out-of-range lab notification, and drug-drug interaction warnings. We performed analyses with each predictor separately, and performed analyses combining them into a composite group of key EHR features. Based on prior work that examined associations between certain EHR features and clinical outcomes,

 we developed a matrix between CDS functions and the clinical outcomes of interest and performed analyses accordingly (

Our main outcomes of interest were based on national quality metrics related to adult primary care: blood pressure control (defined as a systolic blood pressure less than 140 mm Hg), age- and gender-appropriate cancer screening (mammography, Pap smear, and sigmoidoscopy or colonoscopy), health education for particular conditions, influenza vaccination during the months of October through March, and adverse drug events (as measured by visits coded to be related to an adverse drug event). We also combined the receipt of cancer screening, health education, and influenza vaccination into a composite “preventive care” outcome measure.

We were interested in the difference in outcomes based on whether a practice had a CDS function or not. Specifically, we examined whether there were significant differences in the rates of the clinical outcomes depending on whether a visit was to a practice with the CDS or without the CDS.

We categorized practices into 3 groups: all CDS tools active; without 1 or more CDS function; or any disabled CDS. We considered the practice to have all CDS active if they answered “yes” to having all 5 of the individual CDS. We considered the practice to be without 1 or more CDS functions if they answered “no,” “turned off,” don’t know,” or “blank” to any of the CDS functions. We considered the practice to have “disabled CDS” if they answered “turned off” for any of the CDS functions.

We collapsed practice ownership into 5 categories: physician or physician group, community health center, health maintenance organization, medical or academic health center, and other. All other covariates were used as provided by NCHS.

The unit of analysis was the patient visit. We followed NCHS guidelines based on the hierarchical sampling design. We performed all statistical tests on data with less than a 30% relative standard error (the standard error divided by the estimate itself) and more than 30 sample records.

We used multivariable logistic regression modeling to control for a number of patient and practice level covariates. We modeled age as a continuous variable, and all other covariates as categorical variables. We evaluated categorical variables with the χ2 test and continuous variables with the t test. We performed multivariable logistic regression modeling for each quality measure individually. The main predictors of interest were the status of the EHR CDS function. We considered P values of less than .05 to be significant.

We used SAS statistical software (version 9.3; SAS Institute, Cary, North Carolina) for all analyses.

From 2006-2009, the NAMCS/NHAMCS databases included 104,102 adult primary care visits, representing 2 billion adult primary care visits in the United States. Of these, 45% of visits were to practices with EHRs. Of those visits, 97% also had at least 1 of the 5 CDS functions of interest. Of visits using EHRs with at least 1 CDS function, 77% did not have (for any reason), and 15% had disabled (actively turned off), at least 1 of the 5 CDS functions of interest.

We compared patient and practice demographics for visits to clinics with all the CDS functions versus visits to clinics with at least 1 CDS function missing (

). Patients were of similar age (50.9 years vs 51.1 years,

 = 0.84). Rates of visits for flares of chronic problems (7%) and pre- or post surgery (2%) were similar between the 2 groups; rates of visits for new problems (39% vs 41%) and follow-up of chronic problems (32% vs 33%) were less common and visits for preventive care (17% vs 16%) were more common at clinics without at least 1 of the CDS functions versus clinics with all the CDS functions (

 <.0001). Of visits to practices with at least 1 CDS function missing versus those to practices with all the CDS functions, rates of visits to health maintenance organizations (3% vs 7%) were lower, and to physician groups (60% vs 50%), community health centers (5% vs 3%), or medical/academic health centers (18% vs 12%) were higher (

 <.0001). There were no differences in the distribution of visits to clinics without at least 1 of the CDS functions versus visits to clinics with all the CDS functions by gender, race, or region of the country.

Most patients who made visits (81%) had controlled blood pressure. Of eligible visits, there were low documented rates of age-appropriate cancer screening (13%) and influenza vaccination (4%). Approximately 0.2% of visits were related to an adverse drug event. Patients who made visits to clinics with all the CDS functions present versus those who visited clinics that were missing at least 1 of the CDS functions were more likely to have controlled blood pressure (86% vs 82%; OR 1.3; 95% CI, 1.1-1.5) and more likely to not have adverse drug event visits (99.9% vs 99.8%; OR 3.0; 95% CI, 1.3-7.3) (

We found similar results when specifically analyzing visits to clinics with all the CDS functions versus visits to clinics with any of the CDS functions disabled. Patients were more likely to have controlled blood pressure (86% vs 83%; OR 1.5; 95% CI, 1.02-2.3) and more likely to not have adverse drug event visits (99.9% vs 99.8%; OR 5.2; 95% CI, 1.4-19.0) (

We next examined only those CDS functions directly related to each clinical outcome. There were higher odds of receiving cancer screening if the visit was to a clinic with electronic lab results (15% vs 10%; OR 1.5; 95% CI, 1.03-2.2) or notification of acceptable lab ranges (16% vs 11%; OR 1.4; 95% CI, 1.03-1.8) than if the visit was to a clinic without these CDS functions. There were higher odds of having controlled blood pressure (86% vs 80%; OR 1.4; 95% CI, 1.3-1.6) if the visit was to a clinic with electronic problem lists rather than to a clinic without this EHR function (

Because of extremely small numbers and/or absence of outcomes in certain subgroups, we were unable to analyze the association of disabling particular CDS functions with specific clinical outcomes.

Using national surveys of ambulatory visits to primary care clinics, we found significant associations between the use of CDS and some, but not all, clinical quality measures before the enactment of meaningful use. Physicians reporting use of CDS had patients with better blood pressure control and fewer visits related to adverse drug events; these associations were present regardless of whether the physicians with CDS were compared with physicians without CDS or with physicians with disabled CDS. Physicians who reported having electronic problem lists had better rates of blood pressure control of their patients. Physicians with electronic lab results or notification of lab ranges were more likely to provide ageappropriate cancer screening. As we enter stages 2 and 3 of meaningful use in which clinicians must use CDS and then provide data about their performance with respect to patient outcomes, hopefully performance will continue to improve, though this remains to be demonstrated.

The incidence of not having at least 1 of the CDS functions was common, which is likely because the meaningful use mandate was not yet in place at the time of these national surveys. However, despite not having this mandate, the incidence of disabling any CDS was infrequent. Keeping decision support active is a “best practice” in clinical decision support,

but prior experience has demonstrated that many practices elected to disable specific types of decision support before meaningful use. These decisions can sometimes be indicated if the EHR is generating too many false positive warnings, which can create alert fatigue, although there typically are better solutions. The low rate of disabling CDS is reassuring and may indicate that the performance characteristics of CDS are improving. Notably, the odds of having controlled blood pressure and avoiding adverse drug events were higher at clinics with active CDS compared with those who had disabled it.

Prior studies of EHR use and clinical quality outcomes have been inconsistent in their findings. One study using earlier data from the same surveys used here found almost no associations between CDS use and clinical quality measures.

A more recent study found a significant, although small, improvement in blood pressure control of patients visiting physicians with CDS functions in their EHR versus patients visiting physicians without EHRs or CDS.

 In our study, we found an association between general CDS use and blood pressure control, and more consistent associations between quality indicators and CDS functions that were clinically related to these measures. These associations may be present because EHRs and CDS are evolving in a positive direction, spurred in part by the requirements of certification and also those of meaningful use.

Despite the positive findings, we have also found a lack of association between general CDS use and cancer screening, influenza vaccination, or health education, just as many previously have shown a lack of association between general EHR use and certain health outcomes.

 These findings may be because it is harder to change outcomes than quality processes. Some diseases, such as diabetes and hypertension, may be more amenable to clinical decision support than others, in part because there are clear guidelines and metrics.

This study has several limitations. We could not directly measure actual use of CDS during a visit, as the survey consisted of self-reported metrics. Even when physician practices have EHR features at their disposal, clinicians find ways to bypass or ignore them.20,21 Of the small number of practices that chose to disable CDS functionality, we do not know the reasoning behind their decisions, which may, in some cases, have been justified. In addition, the distinction between not having CDS functions and having turned off CDS may be sufficiently nuanced as to lead survey respondents to misclassify their EHR, although this would not affect our analysis of the difference between visits to clinics with all 5 of the CDS functions versus those without CDS. In addition, there were low rates of cancer screening and influenza vaccination captured in this visit-based incident data. There were also low rates of adverse drug event outcomes.

In contrast with some other studies, we found that the use of CDS functions was associated with improvements in measures of health quality at a national level. In addition, decisions to disable these CDS functions, although infrequent, negated these increases in care quality. With the continued implementation of meaningful use, there should be a further decline in practices without EHRs or that choose not to implement or to disable CDS. Overall, meaningful use standards that include CDS appear to have a significant positive effect on some national quality-of-care indicators and health outcomes. It will be important to evaluate the evolving impact of meaningful use as the stages continue to be more widely implemented and better integrated with care processes; we anticipate further gains in healthcare quality indicators and outcomes as a result.

 Division of General Medicine and Primary Care, Brigham Women's Hospital (RGM, JAL, DWB, AB); Brigham and Women’s Hospital (RGM, JAL, AB); Harvard Medical School (RGM, JAL, DWB, AB); Harvard Medical School Center for Primary Care (AB); Harvard School of Public Health (DWB), Boston, MA.

 Dr Mishuris is supported by an Institutional National Research Service Award to the Harvard Medical School Fellowship in General Medicine and Primary Care (T32 HP10251). No sponsor had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript; or decision to submit the manuscript for publication.

Dr Bates holds a minority equity position in the privately held company Medicalis, which develops Web-based decision support for radiology test ordering, and he has served as a consultant to Medicalis. Dr Bates serves as an advisor to Calgary Scientific, which makes technologies that enable mobility within electronic health records. He is on the clinical advisory board for Patient Safety Systems, which provides a set of approaches to help hospitals improve safety. He has received funding support from the Massachusetts Technology Consortium. Dr Bitton serves part-time as a senior advisor to the Center for Medicare and Medicaid Innovation (CMMI). The research presented in this paper was conducted solely within the scope of his faculty appointments and research at Brigham and Women’s Hospital and Harvard Medical School, and in no way reflects any official position of CMMI. The authors report no other conflicts of interest.

 Concept and design (RGM, JAL, DWB, AB); acquisition of data (RGM, AB); analysis and interpretation of data (RGM, JAL, DWB, AB); drafting of the manuscript (RGM, JAL, AB); critical revision of the manuscript for important intellectual content (RGM, JAL, DWB, AB); statistical analysis (RGM, JAL, AB); administrative, technical, or logistic support (DWB); and supervision (JAL).

Using EHR clinical decision support is associated with improved quality of care. Most primary care practices are missing at least 1 "meaningful use" clinical decision support module.

Using Electronic Health Record Clinical Decision Support Is Associated With Improved Quality of Care

To evaluate the impact on smoking status documentation of a payer-sponsored pay-for-performance (P4P) incentive that targeted a minority of an integrated healthcare delivery system’s patients.

Three commercial insurers simultaneously adopted P4P incentives to document  smoking status of their members with 3 chronic diseases. The healthcare system responded by adding a smoking status reminder to all patients’electronic health records (EHRs). We measured change in smoking status documentation before (2008-2009) and after (2010-2011) P4P implementation by patient P4P eligibility.

The P4P-eligible patients were compared primarily with a subset of non—P4P-eligible patients who resembled P4P-eligible patients and also with all non–P4P-eligible patients. Multivariate models adjusted for patient and provider characteristics and accounted for provider-level clustering and preimplementation trends.

Documentation increased from 48% of 207,471 patients before P4P to 71% of 227,574 patients after P4P. Improvement  from 56% to 83% occurred among P4P-eligible patients (adjusted odds ratio [AOR], 3.6; 95% confidence interval [CI], 2.9-4.5) and from 56% to 80% among the comparable subset of non—P4P-eligible patients (AOR, 3.0; 95% CI, 2.3-3.9). The difference in improvement between groups was significant (AOR, 1.3; 95% CI, 1.1-1.4; P = .009).

A P4P incentive targeting a minority of a healthcare system’s patients stimulated adoption of a systemwide EHR reminder and improved smoking status documentation overall. Combining a P4P incentive with an EHR reminder might help healthcare systems improve treatment delivery for smokers and meet “meaningful use” standards for EHRs.

Routine documentation of smoking status improves clinicians’ delivery of tobacco treatment and is mandated in “meaningful use” standards for electronic health records butcan be challenging to accomplish.

		A payer-sponsored pay-for-performance incentive prompted systemwide action by the healthcare system, resulting in improved documentation among both targeted and nontargeted patients.

		The effect was greatest among targeted patients, suggesting the financial incentive added value as well as prompting the reminder.

		Combining a performance incentive with an electronic reminder could help healthcare systems to improve treatment for tobacco use and enable population health interventions for smokers.

Smoking kills more than 440,000 people in the United States annually and remains the leading preventable cause of death.

Despite the availability of effective treatments for tobacco dependence, physicians assess smoking at only 63% of visits and  offer counseling to only 21% of smokers.

 Identification and documentation of smoking status are the first steps to addressing tobacco dependence and have been shown to increase physicians’ delivery of treatment.

 Clinical practice guidelines for treating tobacco dependence recognize the importance of documentation and call for routine smoking status screening and documentation by healthcare systems. The US government’s “meaningful use” electronic health record (EHR) incentive program requires smoking status identification in a coded field.

 Yet achieving a high rate of smoking status documentation is a challenge for many healthcare systems.5

Efforts to improve smoking status documentation have included electronic or paper-based reminders, performance feedback, and a simple vital sign stamp.

 Pay-for-performance (P4P) incentives have been used in some systems to promote delivery of guideline-based tobacco treatment, or as a component of broader quality improvement efforts.

 These studies have examined performance incentives for various guideline-based treatment activities, including smoking status documentation, documentation of physician-delivered counseling, referral to telephone counseling, and payments to providers for patients’ tobacco abstinence. Prior studies of P4P for smoking status documentation were limited to US physician groups or healthcare delivery  systems outside the United States.

 Relatively little is known about the effects of P4P programs in large, multipayer, integrated US healthcare delivery systems in which P4P incentives do not apply to all patients in the system. A payer-sponsored P4P incentive in a multipayer system may only be effective for the patient population to whom it applies. However, practice changes stimulated by a targeted P4P measure can have broader benefits that improve documentation and treatment delivery for targeted and non-targeted patients.

A P4P incentive that was introduced in 2010 in a large multipayer healthcare delivery system rewarded practices for documenting the smoking status of patients with specific commercial insurers and 3 chronic diseases. The organization’s multiple payers and diverse provider groups with varied practice styles and patient populations make standardized quality improvement efforts especially challenging. Our objective was to study the effect of a targeted, payer-sponsored P4P incentive payment on smoking status documentation across the healthcare system.

Partners HealthCare Inc is a large integrated healthcare delivery system in eastern Massachusetts whose provider network, Partners Community HealthCare, Inc (PCHI), represents more than 5000 primary care and specialist physicians, and works with multiple payers including commercial insurers, Medicare, and Medicaid.

 All PCHIaffiliated practices use one of several different EHRs. This study was restricted to the practices using the system’s predominant EHR, the Longitudinal Medical Record, a locally developed system used by 63% of PCHI practices that was first implemented in 1998. Active providers have been using the EHR for a median of 3.5 years (interquartile range, 1.4-6.7 years). It includes progress notes, laboratory results, medication lists, electronic prescribing, and a variety of clinical reminders.

Since 2001, PCHI has negotiated P4P contracts with 3 large, commercial, not-for-profit insurers who collectively cover the majority of commercially insured patients.

 Starting on January 1, 2010, the 3 commercial insurers contracted separately with PCHI to pay practices for achieving a target smoking status documentation rate among a group of highrisk patients. The incentive was implemented using a withheld amount that was returned to practices meeting the prespecified target. Selected targets must meet agreed-upon standards of care that are easy to measure, likely to improve quality of care, and involve enough patients for statistical reliability.

 The amount of payment to practices depended on revenue from P4P-eligible patients and ranged from 3% to 4.8% of practice revenue from the participating insurers for all P4P measures, and $3.8 million was at risk in PCHI for the smoking status documentation measure. Based on the preincentive documentation and the national guideline recommending that patients be screened at every visit, the target documentation rate was set at 80%.

Documentation was measured among eligible patients over a 2-year span (January 1, 2010, to December 31, 2011), with payment at the end ofthe first year for progress toward the goal and at the end of the second year for reaching the goal.

18 years old) who made a visit to a PCHI outpatient practice during the measurement period, were insured by 1 of 3 participating commercial insurers, and had a high risk chronic condition (hypertension, diabetes, or coronary heart disease). The eligible visit could have been to any PCHI specialist or primary care practice in academic-affiliated or community-based practices.

To help practices reach the 80% target, PCHI added an organizationwide clinical decision support tool consisting of a clinical reminder to document smoking status in all patients’ EHRs. The nontargeted EHR reminder was implemented concurrently with the P4P program on January 1, 2010. The EHR-based reminder was designed so that clicking the reminder linked to the coded field for smoking status documentation. Documentation could also be accessed in the EHR through a vital signs entry screen or a health monitoring grid that tracks preventive care and chronic disease management.

To measure the effects of the P4P program on prevalent smoking status documentation, we conducted an observational study before and after P4P implementation. We compared smoking status documentation between the group of high-risk patients who were targeted by the P4P incentive and (1) all non—P4P-eligible patients and (2) a subset of non– P4P-eligible patients who most resembled the P4P-eligible group by having commercial insurance and the same targeted diagnoses as P4P-eligible patients. The study was approved by Partners HealthCare System’s Institutional Review Board.

We used data from the EHR to estimate the effect of the P4P incentive on smoking status documentation. We identified adult patients (

18 years old) who had an office visit with a PCHI provider before (2008-2009) or after (2010-2011) the P4P incentive. We extracted patient data including smoking status, age at the visit, sex, race/ethnicity and primary language entered at registration, insurance, primary care proWe identified the 3 high-risk, chronic conditions included in the P4P incentive. Patients were designated as having the chronic condition if the diagnosis was entered into the EHR before the qualifying visit for that period. Hypertension was defined as a coded entry on the problem list, with the last systolic blood pressure greater than 135 mm Hg or the last diastolic blood pressure greater than 85 mm Hg. Coronary heart disease was based on coded problem list entries of coronary arteriosclerosis, angioplasty, stent placement, coronary artery bypass graft, or myocardial infarction. Diabetes was based on a coded problem list entry of diabetes or a glycated hemoglobin value greater than 7.0%.

We included provider-level demographic data (age and sex) for the primary care provider designated in patients’ EHR registration at the start of the study year. Provider-level data were taken from a master provider list maintained by PCHI. Some provider-level data were missing; for these, we included a dummy variable for missing data.

Smoking status documentation is designated in a structured field in the EHR as “active smoker,” “past smoker,” and “never smoker.” To measure the change in documentation before and after implementation of the P4P incentive, we calculated prevalent documentation of smoking status on December 31, 2009, and December 31, 2011. To match the 2-year duration of the P4P program, we pooled patients with at least 1 visit in the 2 years before (2008-2009) and the 2 years after (2010-2011) the P4P implementation. We hypothesized that documentation prevalence would not be significantly different from year to year before P4P implementation, and confirmed this hypothesis prior to pooling.

We calculated the unadjusted prevalence of smoking status documentation before and after the P4P incentive among P4P-eligible patients (as defined above), among all non—P4P-eligible patients, and among a subset of the non– P4P-eligible patients who were comparable to the eligiblepatients in having had a visit to a PCHI provider and a qualifying  high-risk chronic condition, but a nonparticipating commercial insurer. We measured the prevalence of documentation among providers stratified by the proportion of P4P-eligible patients seen.

We used multivariable logistic regression modeling to test the hypothesis that P4P measures plus the EHR reminder increased rates of smoking status documentation after the contract was introduced on January 1, 2010. We compared the P4P-eligible patients with the comparable subset of non—P4P-eligible patients before and after P4P implementation using first-order interaction terms. We adjusted for patient age, sex, race/ethnicity, language, and number of visits per year, and provider age and sex. To account for clustering, we chose to cluster by provider rather than practice as the higher-level unit of analysis because changing provider behaviors was the primary aim of the P4P and the EHR reminder. We accounted for clustering using generalized estimating equation techniques.

We investigated practice-level effects by calculating the proportion of visits to each practice by patients with P4P insurance. We found no association between clinics’ proportion of P4P-insured visits and patient eligibility, so we did not include this in multivariable analyses. We calculated adjusted odds ratios (AORs) comparing smoking status documentation before and after P4P implementation among the P4P-eligible group and the non—P4Peligible comparison group. The first-order interaction terms of time (before and after) by P4P eligibility were used to test whether patients targeted by the P4P incentive had higher rates of documentation than those not targeted. We tested the difference in change in documentation by eligibility. Analyses were performed with SAS software, version 9.3 (SAS Institute Inc, Cary, North Carolina).

From 2008 to 2009, 207,741 adults had 1 or more visits to PCHI outpatient providers who used the predominant EHR system for their records. Among these, 5671 (3%) were eligible for the P4P incentive and 1209 (1%) were in the similar but non—P4P-eligible control group (

). The P4P-eligible and similar but non—P4P-eligible patients together were seen by 807 providers, who individually saw between 1 and 191 of the patients with 1 of the 3 chronic diseases and commercial insurance. In 2010 and 2011, after the P4P introduction, 227,574 adults visited outpatient clinics using the EHR; 10,236 (4%) were P4P eligible and 2120 (1%) were in the non–P4P-eligible comparison group (Figure 1).

Compared with all non-P4P patients, P4P-eligible patients were older and more likely to be male, black, and English speaking, and to make more visits (

). Providers for P4P-eligible patients were more likely to be female and were older. The subset of non—P4P-eligible patients were slightly older (55 vs 54 years) but were otherwise similar to P4P-eligible patients.

Overall, smoking status documentation increased each year among all patients seen during the study period, from 47% in 2008 and 49% in 2009 to 63% in 2010 after the P4P intervention and 74% in 2011. The relative increase in documentation from 2008 to 2011 was the largest among never smokers, with documentation increasing by 26% in this group compared with

an 18% increase in past smokers and a 14% increase in active smokers. By eligibility, the increase in documentation was greater among P4P-eligible patients than among all non—P4P-eligible patients or the subset of non–P4P-eligible patients who were comparable to the P4P-eligible patients (

). When patients were pooled into 2-year periods before and after the P4P intervention, documentation increased among the P4P-eligible patients from 56% before (2008-2009) to 83% after P4P implementation (2010-2011) (

). The increase was similar among the subset of comparable but non—P4P-eligible patients (56% to 80%) and smaller among all non–P4P-eligible patients (48% to 71%).

The mean documentation by provider among those who saw no P4P-eligible patients increased from 30% before P4P implementation (median 20%; range 0%-100%) to 59% after P4P implementation (median 66%; range 0%-100%). Among providers who saw any P4P-eligible patients, the change in mean documentation increased by the proportion of P4P-eligibile patients seen (lowest tertile: 55% before [median 67%; range 1%-100%] to 68% after [median 77%; range 15%-100%]; middle tertile: 52% before [median 50%; range 0%-100%] to 70% after [median 82%; range 13%-100%]; highest tertile: 54% before [median 59%; range 0%-100%] to 74% after [median 84%; range 17-100%]).

In multivariable logistic regression models, both P4Peligible and the similar but non—P4P-eligible patients were more likely to have a documented smoking status after the P4P was implemented. We compared smoking status documentation before and after the P4P was implemented among P4P-eligible patients (AOR 3.6; 95% confidence interval [CI], 2.9-4.5; 

<.001) and the similar subset of non-P4P patients (AOR 3.0; 95% CI, 2.3-3.9; 

 <.001) (Table 3). Before the P4P intervention, there was no difference in documentation between the P4P-eligible patients and the similar subset of non—P4P-eligible patients (AOR1.0; 95% CI, 1.0-1.1; 

= .45). After the P4P intervention, documentation was higher among the P4P-eligible patients compared with the subset of non—P4P-eligible patients (AOR 1.3; 95% CI, 1.1-1.4; 

 = .009). The differencein- differences between P4P-eligible and non—P4P-eligible was significant (

 <.001). Other characteristics that were independently associated with an increase in  moking status documentation were older age, female sex, black or Hispanic race, non—English speaking, having a younger primary care physician, and having a female primary care physician.

A targeted P4P incentive from insurers that stimulated adoption of a systemwide EHR reminder significantly improved smoking status documentation in a large multipayer integrated healthcare delivery system. Although the P4P incentive applied to only a minority of patients in the system, smoking status documentation rates increased among all patients.

Among patients targeted by the P4P incentive, the improvement in documentation could be attributable to the performance incentive, the EHR reminder, or both. That the effect was slight but statistically significant among P4Ptargeted patients compared with nontargeted patients suggests that the financial incentive added to the effect of the EHR reminder alone. Our findings are consistent with prior work in single-payer systems or physician group practices that demonstrated the effectiveness of performance incentives for improving smoking status documentation.

 In our own system, a decision support program for chronic disease management had the greatest improvement in measures that were also incentivized with a P4P contract.

 Our data suggest that combining a P4P incentive with an EHR reminder may be an effective way for integrated healthcare delivery systems to promote provider behavior change and reach meaningful use goals for smoking status documentation. However, the reminders may account for a large share of documentation improvement. A healthcare system could achieve improvement in documentation with a reminder alone.

A unique feature of this study was the opportunity to observe the effect of a P4P incentive on patients in the same practices who were not targeted by the incentive. The increase in smoking status documentation among these patients could be attributed to the EHR reminder that was implemented for all patients. Had the EHR reminder not been implemented, providers might still have responded to the targeted P4P incentive by altering their smoking status documentation practice for all patients, rather than trying to figure out which patients were eligible. While we cannot determine whether this “spillover” effect occurred in this observational study, we suspect this behavior would occur if providers believe a performance target is important.

In general, P4P programs that promote documentation improve process measures but have not demonstrated effects on quality of care or patient outcomes.

 Screening patients for tobacco use has been associated with increased treatment delivery in some settings, but increased cessation rates havenot been demonstrated.

 However, the evidence base linking brief clinician interventions with smoking cessation outcomes is strong.

 Furthermore, smoking status documentation may have additional benefits in that it enables healthcare systems to implement chronic disease management tools to help smokers achieve cessation outside the clinical encounter.

We acknowledge several limitations to this study. First, we cannot exclude secular trends as an explanation for the increase in documentation in 2010. Notably, meaningful use standards for EHRs were announced in 2010,

 which created an incentive for healthcare systems to improve documentation. During the time period of this study, our healthcare system was still engaged in planning activities and had not yet implemented interventions to meet the standard. Second, unmeasured confounders may be associated with specific insurers. Insurers might be targeting their members who smoke in other ways that prompt patients to discuss smoking with their provider, or insurers might be contacting high-volume providers to promote assessment of smoking status. Third, we used EHR data to identify patient diagnoses, whereas the P4P incentive was administered with claims data. Accurate diagnoses depend on providers maintaining up-to-date problem lists and may be underreported in our data. We chose to use EHR data, which were readily available and reflected the information available for clinicians in deciding whether to screen patients. If patients with comorbidity were identified in administrative claims but not in our EHR data, they would have been included in our non—P4P-eligible group. This would have made our groups more similar and biased our results toward no effect. A fourth limitation is that the P4P incentive mandated that smoking status be documented at the visit or at some earlier time. It was unknown whether smoking status entries from the past were still accurate. The increase in smoking status documentation also identified more never smokers than current smokers. While the utility of documenting smoking status lies in the system’s ability to identify a population of smokers who can be targeted for treatment and care management, increasing overall documentation increases accuracy of smoking prevalence estimates and encourages healthcare system action and investment.

In conclusion, a limited P4P incentive targeting only a minority of a healthcare system’s patients stimulated the system to adopt a universal EHR reminder to prompt smoking status documentation. The combination improved smoking status documentation among all patients, although improvement was greater among patients targeted by the P4P incentive than among those who were not. It appears that the P4P incentive added value beyond its prompting of a systemwide EHR reminder. These findings suggest that combining a P4P incentive with an EHR reminder could help healthcare systems striving for meaningful use goals to improve treatment for tobacco use and enable population health interventions for their patients who smoke.

 From Division of General Medicine (GRK, NAR), Department of Epidemiology (YC), Tobacco Research and Treatment Center (JHKK, NAR), Department of Medicine, Massachusetts General Hospital, Boston, MA; Harvard Medical School (GRK, YC, JAL, JSE, NAR), Boston, MA; High Performance Medicine (JHKK), Clinical Informatics Research and Development (JSE), Partners HealthCare, Boston, MA; Division of General Medicine and Primary Care (JAL, JSE), Brigham and Women’s Hospital, Boston, MA; Mongan Institute for Health Policy (NAR), Boston, MA

A payer-sponsored pay-for-performance incentive for smoking status documentation prompted implementation of a systemwide electronic reminder and improved documentation among all healthcare system patients.

Healthcare System Effects of Pay-for-Performance for Smoking Status Documentation

 Adding clinical decision support (CDS) to electronic health records (EHRs) is required under meaningful use legislation, but there has been little national data on effectiveness in improving clinical outcomes. We sought to determine whether EHRs with CDS improved blood pressure control in US primary care visits.

 We used a cross-sectional, nationally representative survey.

 We examined adult visits to primary care physicians using the 2007 and 2008 National Ambulatory Medical Care Survey (NAMCS).

: We found that patients had a mean age of 52 years, 34% were male, 15% had diabetes, and 70% were white. Rates of blood pressure control were significantly higher in visits where both an EHR and CDS (79%) were used, compared with visits where physicians used neither tool (74%; 

	remained higher after adjusting for potential confounders. In unadjusted analyses, mean systolic blood pressure was 2 mm Hg lower in visits with the use of both an EHR and CDS, compared with visits where physicians used neither tool (

Physician use of EHRs with CDS is associated with increased blood pressure control and lower mean systolic blood pressure across US primary care visits.

Jeffrey A. Linder, MD, MPH

Articles

Antibiotic Prescribing for Respiratory Infections at Retail Clinics, Physician Practices, and Emergency Departments

Using Electronic Health Record Clinical Decision Support Is Associated With Improved Quality of Care

Healthcare System Effects of Pay-for-Performance for Smoking Status Documentation

Electronic Health Records, Clinical Decision Support, and Blood Pressure Control