Medical Home Implementation: A Sensemaking Taxonomy of Hard and Soft Best Practices

Abstract

Context

The patient-centered medical home (PCMH) model of care is currently a central focus of U.S. health system reform, but less is known about the model's implementation in the practice of everyday primary care. Understanding its implementation is key to ensuring the approach's continued support and success nationally. This article addresses this gap through a qualitative examination of the best practices associated with PCMH implementation for older adult patients in primary care.

Methods

I used a multicase, comparative study design that relied on a sensemaking approach and fifty-one in-depth interviews with physicians, nurses, and clinic support staff working in six accredited medical homes located in various geographic areas. My emphasis was on gaining descriptive insights into the staff's experiences delivering medical home care to older adult patients in particular and then analyzing how these experiences shaped the staff's thinking, learning, and future actions in implementing medical home care.

Findings

I found two distinct taxonomies of implementation best practices, which I labeled “hard” and “soft” because of their differing emphasis and content. Hard implementation practices are normative activities and structural interventions that align well with existing national standards for medical home care. Soft best practices are more relational in nature and derive from the existing practice social structure and everyday interactions between staff and patients. Currently, external stakeholders are less apt to recognize, encourage, or incentivize soft best practices.

Conclusions

The results suggest that there may be no standardized, one-size-fits-all approach to making medical home implementation work, particularly for special patient populations such as the elderly. My study also raises the issue of broadening current PCMH assessments and reward systems to include implementation practices that contain heavy social and relational components of care, in addition to the emphasis now placed on building structural supports for medical home work. Further study of these softer implementation practices and a continued call for qualitative methodological approaches that gain insight into everyday practice behavior are warranted.

The patient-centered medical home (PCMH) model of care is currently a central focus of U.S. health system reform, with hundreds of PCMH demonstration projects being carried out nationally (NCMHI 2013). The PCMH is viewed as a means to revitalize primary care delivery, stabilize and lower costs within the system, encourage more doctors to choose careers in primary care, and shift the overall system's focus toward prevention and keeping people healthy (McNellis, Genevro, and Meyers 2013; Rittenhouse, Shortell, and Fisher 2009). Payers and employers who are seeking to lower costs while simultaneously improving care quality are investing in the model through new forms of payment, such as prospective care management capitation, and by providing technical and personnel support for practices making the change. The PCMH model consists of several principles: holistic patient care, emphasis on quality and safety, enhanced care coordination and management, team-based care, enhanced access to care, greater patient engagement in care, and enhanced payment (NCMHI 2013).

The National Committee for Quality Assurance (NCQA) has developed a standardized, process-oriented assessment tool to help determine whether primary care providers and their practices are offering medical home care that is in line with these standardized criteria (NCQA 2013). While other standards for judging PCMH care do exist, the NCQA standard has become the most commonly used assessment template for PCMH care in physicians’ practices. Because payers tie enhanced medical home funding to the NCQA's recognition of PCMH, it is difficult for any practice to market itself as a “patient-centered medical home” without showing that it has met the NCQA's standards. Even so, the risks of taking a “one-size-fits-all” approach to PCMH implementation based solely on a standard set of process-oriented criteria are being debated (Hoff 2012; Rittenhouse, Shortell, and Fisher 2009). The chief danger mentioned in the literature is that a standard implementation approach will lead to medical home activities that either are inappropriate or do not produce the desired outcomes for certain populations (Epstein et al. 2010; Hoff 2012).

The Need to Know More about PCMH Implementation at the Practice Level

Implementation of the PCMH model is at a greater risk for failure because it is new (Nembhard et al. 2009). This greater risk is due to the inherent uncertainty associated with innovation and the common inability to identify in advance the critical factors for successful implementation. Currently, only a few studies have looked at how PCMH implementation works on an everyday basis in practices, although some have shed light on the organization of successful PCMH transformation (Bitton et al. 2012; McAllister et al. 2013; McNellis, Genevro, and Meyers 2013; Nutting et al. 2010; Scholle et al. 2013). Among other things, these studies have shown how the different structural and process ingredients, such as leadership, information technology, teamwork, and outside facilitation, play a role in effective PCMH care delivery.

These studies are notable for identifying a heavy contextual influence on PCMH transformation, as ingredients like financial incentives, teamwork, practice leadership, and electronic medical records are made sense of and perform differently across staff and practices (Bitton et al. 2012; McAllister et al. 2013; Scholle et al. 2013). It is this finding in particular that my article builds on, exploring further how medical home staff use sensemaking and “point of interaction” agency when trying to follow the normative guidance provided for PCMH care delivery while at the same time doing what their knowledge and experience tells them is necessary to implement such care. This article also looks more directly at the everyday implementation of PCMH care components for a specific patient cohort, here, older adults. This makes it a valuable addition to the growing medical home literature because few studies have analyzed transformation or implementation for a particular group of patients (for an exception, see Berry et al. 2013), despite the assertion that the PCMH model should, by definition, vary in different situations and settings (see Stange et al. 2010).

Despite this early work, evaluation of the PCMH model continues to lag behind its rapid advocacy and implementation (Barr 2008; Hoff, Weller, and DePuccio 2012; Peikes et al. 2012). Many PCMH demonstrations under way are not set up for adequate evaluation (Bitton, Martin, and Landon 2010). Many medical homes still remain “black boxes” in which implementation best practices remain hidden from clear view, and the wide variety observed in medical home designs and implementation approaches makes cross-site comparison challenging (Bitton et al. 2012; Hoff, Weller, and DePuccio 2012). Both Crabtree and colleagues (2011) and Bitton and colleagues (2012) call for increased emphasis on everyday PCMH implementation experiences, with the data ideally obtained through qualitative, inductive approaches to better understand medical homes at the individual provider and patient levels. My study was conducted in the spirit of this call.

The use of a tool like the NCQA PCMH standardized assessment to guide medical home transformation does have strengths. Among other things, it provides a specific template that identifies the types of actions and processes thought to yield effective medical home implementation and care, which allows for comparisons across practices and a reliable standard of care. The NCQA assessment focuses on infrastructure availability (e.g., Does the practice have an electronic medical record [EMR] and use it for specific tasks?) and the conduct of specific work-related activities (e.g., Does the practice provide summaries of clinical visits to all its patients?). Many areas of the assessment tool do not specify how to implement the necessary actions or processes. But as a fairly specific set of assessment criteria currently used by payers and the government to provide reimbursement and recognition, it directs practices to construct and conduct their medical home care in a way that limits the ways in which they can act.

As a predetermined template that is imposed on PCMH practice organization rather than derived from its everyday implementation realities, the NCQA tool does not easily incorporate the knowledge and experience gained by individual providers, staff, and practices as they attempt each day to make medical home care work. Although one may argue that such a tool is not intended to incorporate the on-the-job learning that comes from medical home implementation, this is the essential limitation of using a static, standardized approach for guidance; it is not equipped to absorb the lessons learned from ongoing implementation experiences. Therefore, this tool may remain only partially correlated with what clinical and nonclinical staff come to see as vital to effective PCMH care delivery.

We know from the innovation literature generally, and health care implementation studies specifically, that organizations process prior knowledge and information in ways that help them navigate new imperatives (Cohen and Levinthal 1990; Nembhard et al. 2009). In addition, conditions in the local organizational and work environments shape implementation, as well as how individuals using an innovation incorporate their everyday implementation experiences into learning and knowledge that shapes future behavior (Rogers 1995). We see this dynamic in health care all the time, such as in the adoption of patient safety initiatives (Koppel et al. 2008), quality improvement efforts (Ferlie and Shortell 2001), and pay-for-performance programs (Rosenthal et al. 2005). The importance of understanding these on-the-ground realities is illustrated in the PCMH case in the recent revelation of the large variations in PCMH design and implementation across practices, despite the NCQA template, with few comparable precedents for how best to make PCMH care work (Hoff, Weller, and DePuccio 2012; Peikes et al. 2012).

Sensemaking, Older Adults, and Medical Home Implementation

Those staff responsible for implementing PCMH care can show us what they believe, based on their experiences and the lessons derived from them, works best in implementing medical home care for specific patient populations. This article focuses on identifying implementation best practices from clinical and nonclinical staff perspectives. Pursuing this sensemaking approach does not deductively impose a standardized PCMH implementation template on the primary care workplace and ask, “What factors will make this template work the same everywhere?”

Instead, a focus on staff sensemaking forces an inductive, real-time assessment of the prospects for a diverse array of PCMH implementation strategies by asking, “What might be different about the actual experience of caring for particular patients (e.g., older adults), and how does that knowledge impact the implementation of PCMH care on an everyday basis?” As stated, medical home care should be tailored somewhat differently across populations (Stange et al. 2010). Thus, the sensemaking approach helps us learn more about how PCMH care delivery might vary across specific patient populations.

A conceptual perspective that centers on understanding providers’ and staffs’ experience differs from two other views that have been predominant in studying effective PCMH implementation. These alternative views are an emphasis on (1) identifying the right mix of financial or economic incentives to drive effective PCMH implementation (Rosenthal 2008) and (2) specifying the infrastructure requirements (e.g., electronic medical records, formal facilitation) that must be in place for effective medical home care (Reid et al. 2010; Rittenhouse, Shortell, and Fisher 2009). Thus, by offering a worker-focused perspective, my article also complements prior work on PCMH implementation that looked at the roles of structure and incentives. A sensemaking view generally highlights how worker agency matters in shaping how larger policies and interventions are carried out in organizations (Weick 1995).

The sensemaking concept has been cultivated in the field of organizational studies (Weick 1995; Weick, Sutcliffe, and Obstfeld 2005). Weick, Sutcliffe, and Obstfeld (2005, 409) asserted that “sensemaking is, importantly, an issue of language, talk, and communication. Situations, organizations, and environments are talked into existence.” Here, however, the sensemaking perspective assumes that new knowledge and learning about “how things actually work” and “how things should work” with respect to PCMH implementation come directly from the primary care practice staff as the social actors who have interpreted (i.e., “made sense of”) their experiences caring for older adult patients over time.

These experiences may be shaped by a variety of factors: the surrounding everyday practice context, larger primary care practice structures such as payment systems and workforce shortages, medical home imperatives, and the unique characteristics of a given patient population. The new knowledge and learning gained from these experiences subsequently influence how these staff strategize, approach, and act toward medical home implementation. In short, the knowledge and learning help create a perceived reality from which individuals form the basis for future behavior (Laroche 1995; Weick, Sutcliffe, and Obstfeld 2005). Thus, PCMH implementation best practices are uncovered through trial and error and from ongoing engagement with work meant to provide older patients with medical home care. Because of their more emergent nature, such best practices may not align fully with the standard templates of external bodies like the NCQA.

Older adult patients are a relevant population to consider for justifying a sensemaking approach to PCMH implementation best practices. On average, they differ from other groups of patients in key ways. Older adults account for a significant number of primary care office visits in the United States and physicians’ office visits generally (Cherry, Woodwell, and Rechsteiner 2007). Indeed, persons older than sixty-five see their providers two to three times more often than younger individuals do. Older adults also may have unique needs, preferences, and expectations that impact PCMH implementation. For example, since older adults have higher rates of chronic disease and tend to have multiple diseases concurrently, care coordination and case management can become so complex for primary care providers that communicating with several specialists and assembling needed clinical information likely require adjustment at the practice level. In addition, many older patients require ongoing “technical support” from providers in order to feel confident that they understand both their diagnoses and prescribed treatments and their own role in managing them.

Older adult patients also tend to value direct, face-to-face involvement with the same providers (Safran 2003). A higher percentage of older patients are accustomed to the traditional office visit model of care and having one doctor care for them. Consequently, they may prefer a more paternalistic interaction with the system, in which they are comfortable taking orders from a clinician and being told what to do. In this way, older age has been associated with less preference for being proactive in and having control over one's own medical decision making (Levinson et al. 2005). In addition, older patients are more likely than other patients to have functional limitations. This also may undermine their ability to engage in their own care and may reduce the value of technology as a tool for this engagement. Greater functional impairment generally lessens older adults’ use of computers (Carpenter and Buday 2007).

This article provides an opportunity to gain a better understanding of how medical home principles should be implemented “on the ground” for different subpopulations. My analysis was based on two general research questions:

1. What implementation best practices did staff at the practice level identify for older adults and medical home care?

2. How do these implementation best practices compare in form and substance?

Data and Methods

This article is a qualitative study, part of a larger Agency for Healthcare Research and Quality (AHRQ) grant to examine PCMH implementation for older adult patients at the individual provider and staff levels. The data presented here come from fifty-one semistructured interviews with individuals working across six different NCQA-recognized PCMH primary care practices that had been participating for between one and three years in a large PCMH demonstration project carried out by a local insurer (see table 1). These interviews were conducted between November 2011 and October 2012.

TABLE 1.

Interview Sample (n = 51) and Dimensions Associated with the Six PCMH Study Practices

Practice 1 (66% provider staff)a	Practice 2 (60% provider staff)	Practice 3 (46% provider staff)
9 interviews (4 providers, 4 nurses, 1 practice manager)	7 interviews (3 providers, 2 nurses, 1 secretary, 1 practice manager)	10 interviews (6 providers, 2 nurses, 1 care coordinator, 1 practice manager)
Recent level 3 PCMH	Recent level 3 PCMH	Level 3 PCMH
Suburban clientele	Urban, poorer clientele	Suburban clientele
Family medicine only	Internal medicine only	Family medicine only
Practice 4 (50% provider staff)	Practice 5 (46% provider staff)	Practice 6 (100% provider staff)
6 interviews (2 providers, 3 nurses, 1 practice manager)	8 interviews (5 providers, 2 nurses, 1 practice manager)	5 interviews (2 providers [1 twice], 1 nurse, 1 practice manager)
Level 2 PCMH	Recent level 3 PCMH	Recent level 3 PCMH
Rural clientele	Urban, poorer clientele	Rural/suburban clientele
Family medicine only	Family medicine only	Family medicine only
Other interviewees:
1. Two interviews each with the clinical quality supervisor for the primary care network and the clinical quality analyst for the network (n = 4).
2. Interviews with one practice manager and one physician from a seventh PCMH level 3 practice (physician also oversees PCMH implementation for network) (n = 2).

Notes: All practices included in the study were part of a larger organization that helped manage the administrative and staffing requirements for each practice, although each practice was fiscally responsible for funding all its office staff and administrative demands. Physicians and staff in the practice worked directly for this larger umbrella organization, although the doctors as a whole were its ultimate controlling and decision-making group. In this way, the practices were members of an “independent practice association” type of organization.

^aThis number is the percentage of total provider staff (i.e., physicians, nurses, and physician assistants) interviewed in a practice.

The demonstration project offered greater reimbursement (both capitated care management fees per member per month and higher fee-for-service payment for some services) to practices that reached level 3 NCQA-PCMH recognition and specific quality improvement and utilization reduction targets, such as fewer emergency department visits for practice patients, fewer hospital admissions and readmissions, and NCQA-HEDIS targets in specific areas such as chronic disease management. Level 3 accreditation represents those medical homes most advanced in the number of NCQA PCMH assessment criteria they have met. This study was approved by the Northeastern University Institutional Review Board.

Study Approach

Because sensemaking focuses on people's subjective interpretations of their experiences, the methodological approach employed in this study is inductive and qualitative. This approach currently is advocated for studies of medical home implementation (Bitton et al. 2012; Crabtree et al. 2011). It uses primary data collected directly from primary care providers, nurses, and nonclinical staff that treat older adult patients in their practices. Collecting primary data through a qualitative approach is favored when seeking new understanding, when that understanding is best found in the everyday workplace, and when detailed description is preferred for specifying implementation dynamics that require further testing (Miles and Huberman 1994). This makes the study notable, as most of the existing examinations of patient-centered medical homes are quantitative or survey based and take an economic or infrastructure-focused approach, viewing PCMH success as deriving more from the appropriate design of specific reimbursement mechanisms, financial incentives, and practice infrastructure components (cf. Reid et al. 2010; Rittenhouse, Shortell, and Fisher 2009).

The goal of this study was to allow provider and nonprovider staff implementing PCMH for a specific population (i.e., older adults) to provide a window into their experiences, describing how these experiences have shaped their thinking and learning about PCMH care and implementation. As noted earlier, the study is driven by the notion that there is not a “one-size-fits-all” approach to PCMH implementation (Hoff 2012; Stange et al. 2010). This justifies my core focus on gaining idiographic knowledge from practice staff implementing PCMH care for a single patient cohort—older adults.

Data Collection and Analysis

The participants in the study were twenty-two primary care physicians (family physicians or general internists), five nurse practitioners or physicians’ assistants, seven nonclinical practice managers, six registered nurses (RNs), seven licensed practice nurses (LPNs), three medical assistants, and two network-level clinical quality supervisors (one RN and one nonclinical staff person). Having a diverse study population was advantageous, particularly because I was using a subjectively oriented approach (i.e., sensemaking) to guide my data collection and analysis. It allowed an ongoing triangulation of responses across different worker groups while enabling me to compare their opinions, beliefs, and experiences.

Each interview lasted, on average, between forty-five and sixty minutes. Although I conducted most of the interviews in person, I had to do several by phone to accommodate physicians’ and nurses’ schedules. The interviews were guided by a semistructured protocol that contained several “grand tour” questions and smaller probe questions that delved deeper into particular topics, mainly to ensure that the interviewees covered specific aspects of their PCMH implementation experiences. Grand tour questions generally are designed to elicit a wide range of responses by an individual to a particular topic, thus encouraging spontaneity. This spontaneity is critical to anticipate when attempting to gain insight into a topic about which less is currently known, such as PCMH implementation. I conducted all fifty-one interviews, with assistance from a second researcher on fifteen.

The goals of the interviews were (1) gaining descriptions of everyday PCMH implementation experiences from staff, with special emphasis on their older adult patient populations regarding their work improving care coordination and management, access to care, patient self-management, quality and safety, and holistic care; (2) having staff interpret and make sense of what they had learned (e.g., knowledge, beliefs, competencies, skills) from implementing PCMH care for older adults; (3) gaining insight into the types of implementation experiences that the participants viewed as most successful and challenging; and (4) identifying specific organizational (e.g., structural, cultural, human capital) barriers and drivers for effectively pursuing one or more implementation practices.

All the interviews were digitally recorded and transcribed into text documents, which were then analyzed using Atlas.ti (version 6.2) qualitative analysis software. Because of its emphasis on affixing codes and labels to qualitative data (a process called chunking), this software helps researchers develop detailed, analytical, and broad interpretations of the data. It facilitates the types of approaches employed during data analysis for this study, in particular taxonomic analysis (see Spradley 1979). Taxonomic analysis sorts and classifies qualitative data into particular categories of meaning that can be interpreted more specifically. It is used when one of the analytic goals is to develop lists of phenomena that in some cases may have similar characteristics and in others are substantively different. In my study, the categories of meaning or lists desired were different types of PCMH implementation best practices, as articulated through the PCMH practice staff's experiences and sensemaking.

I analyzed the data for this study in the following way. First, I read ten interviews without coding them using Atlas.ti. This initial review helped me become familiar with the major content areas emerging across interviews, for example, the particular types of implementation experiences that the interviewees emphasized. The notes that I took on these ten interviews were reviewed, and these notes served as a tool for performing the initial coding of the first batch of interviews. I then coded approximately five to seven interviews at a time using Atlas.ti. These batches were structured according to individual primary care practice; that is, all the participant interviews in one practice were analyzed in one batch, with the next batch consisting of a new set of practice staff. This facilitated the comparative case study approach. Starting out, it was not known which specific implementation best practices might emerge from the interview data. Because of this, a process known as theoretical sampling was implicitly employed as part of the analytic strategy. Theoretical sampling uses using the information gained from earlier data collection to help direct and focus future data collection (Strauss 1987).

The coding process within each batch of five to seven interviews proceeded in a similar manner. First, open coding was performed where phrases and paragraphs for each interview were labeled in terms of some interpretive focus (e.g., a single implementation experience, best practice, or barrier). These open codes were then organized into lists and aggregated into higher-order codes that captured larger meaning categories (e.g., a category of similar implementation best practices such as “protocol-driven activities” or “family focused behavior”) across interviews within the given batch. As I analyzed each additional batch, I compared the codes with those of earlier batches to determine which types of code categories were consistent and most widely supported across batches.

To determine the level of code support, I assessed the frequency of a particular code both within and across the different interviews, as well as how “rich” the code seemed in regard to its descriptive content. Codes that were described consistently with the same level of detail across participants and also had the support of a majority of interviewees were considered moderately or strongly supported. For this study, two large implementation domains or best practice lists emerged. These were labeled “soft” and “hard” PCMH implementation best practices. Within each of these two domains were lists of individual best practices that each defined a particular type of implementation practice within that particular domain.

Results

My analysis of practice staff interviews yielded two larger domains of best practices, “hard PCMH implementation practices” and “soft PCMH implementation practices.” “Hard” implementation best practices are more standardized in their execution by staff. They involve clearly measurable or assessed (e.g., quantified) activities, are geared explicitly to specific outcomes such as improved efficiency; tend toward work flow and job improvement; are reproducible across settings through protocols and instructions; and are task oriented and transactional in their focus. Because of their transparent nature, hard implementation practices also are acknowledged by external (i.e., outside the practice) stakeholders, such as accrediting bodies, making them normative or expected aspects of practice behavior in regard to medical home implementation (see table 2). Hard practices relied more upon explicit knowledge embedded within the practice, such as within formal policies or routines clearly understood by all staff.

TABLE 2.

Results of Taxonomic Analysis for PCMH Implementation Best Practices (n = 51 staff interviewed)

Best Practice	Description/Examples	Key Ingredients	Latent Function(s)	By-products (+/−)	Evidence as a Perceived Best Practice
Soft (Domain)a
Knowing the patient	Relational/interpersonal: Having an established relationship with patient; knowing their personal situation well	Time, provider interest, staff interest, history with patient, EMR capability	Enhances a strategic approach to individual patient care (personalized care); builds trust in patient-practice relationship	Positive: Improves diagnostic accuracy, patient compliance, and patient satisfaction; can improve scheduling and transactional aspects of practice work.Negative: Can be inefficient and a “time sink,” especially for providers; can place heavier burden on select support staff	Very Strong
				Negative: Can be inefficient and a “time sink,” especially for providers; can place heavier burden on select support staff
Empathy and compassion when dealing with patients	Relational/interpersonal: Identifying with patient's situation and using patient situation on a case-by-case basis when deciding how to approach care/interaction	Time, provider interest, staff interest, history with patient	Builds trust in patient-practice relationship; enhances personalized care	Positive: Improves care management, patients’ compliance, information flow from patient	Strong
				Negative: Can be inefficient and a “time sink” for practice; can place heavier burden on select support staff who “know” patients
Using family	Relational/interpersonal: Practice staff and providers making a concerted effort to involve family members in care when deemed necessary, e.g., informed consent, patient decision making, care transitions	Time, provider interest, staff interest, history with patient and family, family interest, EMR capability	Provides independent verification of patients’ needs and behaviors; enhances personalized care	Positive: Improves care management, patient compliance, diagnostic accuracy	Strong
				Negative: Can hinder care management if too many family members involved; can be inefficient and a “time sink”
Tailored communication approaches	Relational/interpersonal: Relying on personal knowledge of patients and relational elements to tailor communications with them in ways to which they will respond positively	Provider interest, staff interest, history with patient, well-trained staff	Enhances personalized care; makes staff feel better about getting through to patients about orders, etc.	Positive: Improves care management, patients’ compliance; can make visits more efficient and predictable	Moderate
				Negative: Can be inefficient and a “time sink”; can undermine use of EMRs and e-portals; heavier burden on select support staff who “know” patients
Hard (Domain)b
Use of formal work protocols	Focused on work flow: Standardized improvements to work flow characterized by routine process implementation, e.g., medication reconciliation, discharge planning, orders tracking, clinical visit summaries	Motivated practice leadership, EMR capability, integration and relationships with other providers	Gets practice staff “on same page” regarding how to move through various primary care transactions; allows lower-level staff to do higher-level work	Positive: Can make visits more efficient and predictable; increases certainty and information about a patient's situation	Strong
				Negative: Can overemphasize standardization in situations in which contingencies are needed; can undermine relational aspects of care
Work redistribution	Focused on human capital: Moving work from one staff person to another, e.g., nurses doing formal triage and medication reconciliation; front-desk staff doing orders tracking and care coordination	Time, history with patient, relational bond between doctor and patient, well-trained staff, EMR capability, adequate staffing	Can increase overall capacity for serving patients in practice; uses full range of staff skills/training, increasing practice's collective knowledge	Positive: Can make visits more efficient; increases certainty and information about a patient's situation	Moderate
				Negative: Can distance the provider from the patient; potential “voltage” drop of patient information across staff; potentially lower quality in task completion
Triage	Focused on work flow: Having in place a formal or informal system of assessing patients’ situations, e.g., staff conducting phone triage with patients, staff contacting patients after hospital discharge to discuss care needs	Time, history with patient, well-trained staff, EMR capability, relational bond between triage staff and patient	Enhances a strategic approach to individual patient care (personalized care); can increase overall capacity for serving patients in the practice	Positive: Produces more cost-effective care delivery; makes better use of provider's time	Moderate
				Negative: Can distance the provider from the patient; may undermine patient's trust in practice

Notes

^a“Soft” implementation best practices are those that involve heavy relational aspects of care, that are less easily measured or assessed (e.g., quantified), that derive specifically from ongoing social interactions between staff and patient, and that produce intangible by-products or benefits that are best understood directly by those workers actually delivering the care.

^b“Hard” implementation best practices are those that involve more easily measured or assessed (e.g., quantified) activities, that are geared explicitly to focused outcomes such as improved efficiency, that tend toward standardized changes in work flow and job improvement, that are more easily generalizable or reproduced across settings, and whose focus is task oriented and transactional.

The strongest support was for a subcategory of hard best practices that staff identified as “using formal work protocols” to deliver PCMH care to older adults (table 2). These protocols were often written down, implemented as practice policy, and known by all staff in the practice. For staff, the use of protocols enhanced certainty with respect to caring for older patients. Examples included medication reconciliation, discharge planning, orders tracking, and clinical visit summaries. Some practices had begun using these protocols in direct response to the NCQA's medical home accreditation requirements. Others had begun using them to solve a particular care-related or work-flow problem, in a more ad hoc manner. The staff believed that using targeted work protocols, especially for older patients, streamlined treatment and decision making, largely because they facilitated two important goals in their minds: (1) the establishment of a reliable information pool for older patients who often were hard to get information from and had ambiguous needs, and (2) greater accountability for older adults who preferred being told what to do and whose functional limitations might impact their self-care (see table 3).

TABLE 3.

Representative Quotations Illuminating the PCMH Best Practices Identified in the Study

Soft Practices
Knowing the patient	I know this patient; over the course of time we've developed a good relationship. Like another time he came in last year and his complaint was that he was more tired than usual. But then you look at any guy in his eighties, and you think, “Well, I mean you're eighty, what do you expect?” But I knew for a fact that he exercised every day. He competitively swims. He was complaining that he didn't win his last swim meet. He races other seniors. So he said, “I'm slowing down; I just feel kind of tired.” And because I knew this I decided to order an EKG, thinking maybe it's his heart. He had like a first-degree [avioventricular] block. They immediately hospitalized him and gave him a pacemaker. If I didn't get the EKG he would have been dead in two days. That's what the cardiologist told him (physician, practice 1).
Empathy and compassion when dealing with patients	It's all about need and just emotional support. Making the staff understand not to get frustrated with these people. I tell them, “You have to understand the elderly. They don't feel good. They're frustrated. They don't like using voice mail” (practice manager, practice 2).
	I picture my mom [when talking to an elderly patient on the phone]. She's eighty-five, so I talk to them like I would talk to my mother, only nicer. Give you an example—we have a cute little old lady. She just calls and says, “Hey, it's me,” and I know who she is right away. She was given a new medication. I'm like, “Oh, you'll be fine. My mom takes it.” She called back later and said, “Thank you for prescribing that for me.” OK, I didn't prescribe it, but the approach works well. Just do little things. Be nice to them. They just like you to take time and make them feel like they're the most important right then and there (secretary, practice 2).
Using family	I had a patient the other day who was here because she was worried about her blood pressure. When I went through her medications, she didn't know what she was taking, and, come to find out, she wasn't taking any of her medications. She didn't understand what she needed to do. So we called her family member—we were on the phone for a good hour and a half throughout the day, you know, talking back and forth with the daughter. So she's [the daughter] going to get the pills together. The patient's coming back in three weeks for a blood pressure check. The daughter's coming with her (nurse practitioner, practice 2).
Tailored communication approaches	I think that you sort of have to get to their level. I don't mean to sound degrading, but things you have to do with children, as you get older you have memory loss, forgetfulness, you don't understand things as well (nurse, practice 1).

Hard Practices
Using formal work protocols	We started doing preplanning where you call patients and ask them a series of questions before they come in for their annual physical. “Have you been in the hospital recently?” You'd be amazed at how many people have been in the hospital. “Have you been in the ER recently?” and we go through a form that has things like the last time they had their lipids checked, and so when the physician has this information and the patient comes in, he can go through and review and see exactly what they are due to have—if a female was due for a mammogram, a dexa scan, whatever (practice manager, practice 2).
Work redistribution/off-loading	Some of the things that we've also done is try to realign the front-end staff, the secretaries, to use some anticipatory guidance as to what the patients are going to need. All of the paperwork that the federal regulations require us to have done; all of the HIPAA paperwork and consents, all of that stuff can be done with the front-end staff and the medical assistants can then do what they need to do. So some of that medical home concept has pushed us to use people where we need to and when we do that, then the staff sort of take a more active role in owning the process (nurse practitioner, practice 2).
Use of triage	We have some nurses that are trained in triage, and they'll get a phone call about a respiratory tract infection, for example, and they look at the patient's chart in terms of whether the patient has chronic lung disease or sick with any other ongoing medical problems, and they'll make a clinical judgment that the patient's probably not sick enough to be seen, probably has a viral illness, and they'll recommend, “Just try this cough syrup and some antihistamine and watch for a temperature and take some Tylenol and call us if you're getting worse.” And they'll document that on the chart, send us a note about it, and we just say, “Agree with plan,” that shows that we saw it, and it becomes part of the patient's record (physician, practice 3).

Every staff member in the six practices felt that the protocols they used made visits with older patients more efficient and predictable. First, they provided staff with a consistent understanding of older adults’ many and changing needs. A good example is the use of a process by all the practices to help reconcile medications for elderly patients. While a few practices performed this reconciliation over the phone with all patients scheduled for an office visit, several practices had modified this protocol for their older patients, reconciling the medications in person with the patient.

We do medication reconciliation, which is a monstrous piece with the elderly, because they're on so many meds. The patient checks in. We give them a piece of paper that has all of their medications preprinted on it. There is a cover sheet that goes with it giving them instructions. And while they are sitting in the waiting room, they are instructed to cross out, write whatever they want to essentially correct the list. When we sit down with them, we can see if they crossed out anything, like their diuretic, for example. And we can say, “What happened to that?” And they can say, “I haven't been on that for months!” We can say, “Why not?” (physician, practice 2)

We tried to do it [medication reconciliation] by phone, like we were told to do by the national consultant. But it doesn't work with the elderly. You need to have them bring in all of their meds, which we do, in a plastic or brown bag, and then do it on site with them before they see the doctor. (nurse, practice 4)

In addition to the work-flow and decision-making advantages gained by the practice, the use of formal protocols, when tailored to the specific needs and realities of elderly patients, made staff feel that they had better experiences with older patients complying with practice instructions. In this way, they believed that formalizing certain work practices improved the quality of care and empowered many of these patients.

I think the visit summary we give them is really effective. I have noticed a decrease in callbacks after the patient has left. Questioning what it is I wanted them to do. There's less often they're coming back and not having initiated the medication changes that I've instructed. Because even if they don't absorb it, they finally have a piece of paper they can show their family member and say, this is what Dr. Smith gave me. (physician, practice 2)

Another hard implementation practice that received support was the “redistribution of work” from one group of practice staff to another. Staff, mainly physicians, believed that a practice's capacity expanded when individuals were pushed to perform work that extended their skills and capabilities and was based on maximizing the use of all human capital working in a practice, both clinical and nonclinical personnel (table 2). The implementation of this best practice was risky because of the uncertain ability of some staff to perform new types of job duties and responsibilities competently, since most of the work redistribution was to lower-level staff now doing previously higher-level work.

Staff described their experiences and gave examples of work redistribution that moved work usually done by clinical providers to the nursing staff, and work done by the nursing staff to medical assistants and medical secretaries. In this way, the typical redistribution moved down the practice staff hierarchy vertically rather than horizontally and was controlled directly by physicians in the practice, even though many of these physicians presented the process in an egalitarian manner during interviews. Actually, however, no work was redistributed in any of the six practices without the physicians’ either explicit or tacit approval. Examples of redistribution were registered nurses taking on additional patient triage and rooming responsibilities, LPNs and medical assistants overseeing key aspects of medication reconciliations, and front-desk staff tracking test and lab orders.

We have order reports.1 We go in on a monthly basis and pull out the orders to make sure that if I sent you to an eye specialist that you actually followed up. If I don't have the records, I'll call the eye doctor. They'll say, “Well, she never came to the appointment.” Then I call the patient, find out, and get them rescheduled to go. (secretary, practice 2)

Surprisingly, both clinical and nonclinical staff focused less on the potential risk of giving lower-level staff more difficult work to do. The interviews made clear that any perceived risk in this regard was outweighed by a felt need for each practice to gain additional capacity for providers at all costs, especially in the pursuit of medical home goals that involved making office visits more accessible to patients, increasing communication between the practice and its patients, and coordinating care with outside clinicians and facilities.

To many staff, it was the older patient cohort in particular that drained practice capacity with respect to several medical home imperatives. Each practice expressed feelings of being under intense pressure to free up its providers to see more patients and offer more complex care, which usually applied to disproportionately high numbers of elderly patients. Like the use of tailored work protocols, staff believed that this practice helped increase both the efficiency and the predictability of the older patients’ visits. They also believed from their experiences that it did enhance physicians’ and nurses’ capacity to carry out more complicated transactions for those older patients who required them.

Many staff implicitly believed that the practice of redistributing work pushed individuals to perform the way they were capable of performing. This belief was relayed through different experiences staff described that involved their observations or direct participation in a given work redistribution situation. In this way, most practices conveyed a sense that staff, from nurses to receptionists, generally could take on more complex work as a result of the medical home model. Whether or not this was true, it was a strong belief that personnel at all levels often expressed during the interviews.

We do all this prework for physicals and things like that. And now I have a chance to become less of a data entry person. The prework stuff is all done ahead of time now for a good percent of the people. And we have someone who does that job. She used to be our transcriptionist, and she still does a little of that, but most of what she does is take the forms from patients and enter all the information on diagnosis and family history into the medical record. So then I just have to pull it up and look at it. I'm not the one also putting it in. And it just helps with work flow. (physician, practice 3)

The staff felt that the “use of formal triage” was another effectively implemented best practice that increased their capacity to see patients, especially older patients, while providing more personalized care to individuals (table 2). Triage was generally defined as the previsit assessment of patients, over the phone and sometimes in person, who did not yet have a formal visit appointment but had contacted the practice seeking help or a consultation. After using ad hoc triage approaches for several years, all six practices in the study had formalized their triage approach to all patients. This formalization included writing up standard procedures for a various triage situations, dedicating specific practice staff as triage personnel, writing formal job descriptions for them, and building into the workday employee time specifically for triage. Initially, the triage best practice required additional resources because of the additional time needed for staff.

All staff members felt that a well-functioning triage system could help increase providers’ available time for complex older adult patients. In their experience, the triage of older adults by nursing and medical assistant staff added a critical buffer between the healthier, higher-maintenance older patients and the providers. Physicians liked this best practice because cases were identified in advance, with others, such as nurses, doing the initial sorting (table 3). Not only did it make the physicians’ workday easier, but it also appeared to enhance their job satisfaction.

Nurses praised this practice because it was yet another opportunity to do complex work and raise their stature with patients. The triage best practice also offered several advantages to older patients by giving them a less expensive and lower “hassle” outlet for their concerns as well as increasing overall practice goodwill toward them. This extra goodwill resulted from the staff's feeling that they now had greater control over deciding how an elderly patient would gain access to the physician, giving each practice greater predictive power in scheduling visits and managing daily workloads.

Soft PCMH Implementation Best Practices

The second major sensemaking domain that I identified was labeled “soft PCMH implementation practices,” defined as those best practices that involve heavy relational aspects of care (e.g., between staff and patients), that are less easily measured or assessed (e.g., quantified), that derive specifically from ongoing social interaction between staff and patient, and that produce intangible by-products or benefits that are best understood by those workers delivering the care. Soft practices tend to be more covert or hidden from view because they are embedded within the practice's social structure. Soft best practices are also oriented toward informality and reliance on tacit knowledge existing within practices, through a collective mindset among practice staff about the proven (through experience) value of particular ways of communicating and engaging with their older patients. The best practices in this domain are activities and approaches not necessarily articulated within formal PCMH model components or assessments. As a result, such practices may exist at the practice level as meaningful but unrewarded or unrecognized PCMH implementation best practices for older patients. This reality places them at the individual mercies of each practice's ability and motivation to pursue them.

The most important soft best practice identified through staff discussions of their experiences implementing PCMH care for older adult patient populations was categorized as “knowing the patient” (table 2). Staff believed strongly that for elderly patients, this best practice facilitated such hard implementation practices as triage, as well as the other soft best practices identified, making it the main contributor to the PCMH's implementation success for older adults in each of the six practices. Staff defined “knowing the patient” as having a close, detailed understanding of individual older adult patients: their life and living situations; personal, clinical, and emotional needs; and family situations. Staff believed that such an understanding came from extended relationships with their older patients over time, cultivated and maintained through ongoing, face-to-face interactions.

I came to this practice because when I worked in the emergency department, every time I called the physicians [at the practice] and said, “You know, I have John Doe here,” one of these doctors would say, “Oh, you know he's had a tough time of it. He lives with his wife but he's got a couple of kids who live out of town and he's on da, da, da. And I just saw him a couple of days ago.” You know, they had a sense about what's going on with their patients and they're good at knowing everything about their patients. (nurse practitioner, practice 5)

Staff across the six practices emphasized that while knowing each patient well had advantages, it was important that they had deep relationships with older patients. Deep relationships were those with an intimate, rich knowledge of the patient's preferences, limitations, and clinical issues. In the staff's experience, older adults were much more complex than younger patients. They also were at a higher risk for a variety of health-affecting events, ranging from heart attacks to depression to falls. Finally, the staff of the six practices believed that older patients were better at compliance and prevention when they had a close relationship with the practice and the providers and, in the staff's experience, knowing the patient well led to much of this intimacy.

Staff described other tangible benefits of knowing the patient in their everyday implementation of medical home care. First, they believed that it improved older patients’ satisfaction with their provider, the practice, and their overall care.

Older patients want to know a face. I think they come in here and you know, three or four years ago when they came in here and saw a different face every time, I think they got uncomfortable with that, and some of them left the practice. I can't tell you for sure, but I think they like the idea of “My doctor used to come to my house. My doctor delivered my baby and my doctor did all this stuff.” I think that's very important. (physician's assistant, practice 3)

Staff talked about many older patients’ need to rely on their physicians and the practice generally for more than just timely, efficient care. They described experiences interacting with older patients who saw their interactions with the practice as the main social event of their week, a chance to get out of the house and feel connected to the world; who relied on the staff for advice and comfort in dealing with transformative events such as the death of a loved one, especially for those living alone; and who wanted their physicians to tell them specifically what they should do to make their lives better. The staff believed that they could fulfill these needs only with a high level of trust between the older patient and the staff, and such trust came from building an ongoing interactive relationship over time.

From the staff's perspective, knowing the patient increased the practice's efficiency and quality of care. On the efficiency front, staff described many instances in which having in-depth knowledge of an older patient's needs, personality, or other characteristics enabled quicker treatment decisions to be made and personalized older adult care to be carried out in a manner that helped with scheduling visits and managing work flow. The disadvantages of knowing the patients well as an implementation best practice is that it required large amounts of time and placed a heavier burden on those staff in the practice who knew the older patients well and thus were often the ones interacting with them the most. Having enough interactions with an older patient to create mutual trust and to gain insight into the patient's personal life had up-front costs. In these six practices, these costs fell disproportionately on specific staff, particularly those who had worked there the longest. These staff had the most tacit knowledge of particular patients and, if they left, such knowledge could not be easily replaced.

“Empathy and compassion in dealings with older patients,” too, was an important and well-supported implementation best practice (tables 2 and 3). Again, staff believed that this practice, while benefiting patients of many ages, was most important to their elderly clientele, given their higher percentage of complex, often fragile emotional states; their greater life and health uncertainties; and their greater likelihood of more severe clinical conditions. This best practice depended on the staff's knowing a patient well, since the empathy and compassion had to be tailored to the patient from knowledge of his or her circumstances and personality.

We had one of our older patients not that long ago walk in and he had hand shingles. But we know that he also has a ninety-year-old wife at home with dementia that he cares for, and it just happened that the aide was there. So he had time to walk over here [to the practice] because he doesn't live too far away, and he wanted to be seen to check it out. You have to put yourself in their shoes to understand that what might seem like an inconvenience to us because he's walking in and wants to be seen, well he's walking in because of his circumstances, and it's just sad. So, it's a matter of knowing how to deal with them and being accommodating. (nurse, practice 5)

“Using family” in key ways when caring for patients was a third soft implementation practice unique to elderly patients in each of the six practices. This is a soft practice because it, too, required an understanding of the patients and their situations that could best be acquired through extended social interactions that built trust between patient and practice. This best practice had both clinical and nonclinical practice staff finding specific information about an elderly patient's family members and, over time, building a rapport with those family members who, in the staff's minds, were available, willing to help if needed, and comfortable in advocating for their loved ones.

Staff reflected on the initial process of using family in PCMH implementation. For them, it began with accumulating reliable information from the patient, moved into codification of that information in the electronic medical record (making it available to all staff in the practice), and culminated in the actual involvement of one or more family members in a specific care situation. In the staff's experiences, this involvement meant obtaining informed consent for impaired elderly patients needing services, acquiring information about the elderly patient's compliance with care directives, getting the family's input on actual care decisions, using the family as a communication liaison for elderly patients, and having family members assist in care transitions.

Practice staff had mixed feelings about this best practice. On the one hand, they all believed that it improved both the quality and management of the elderly patients’ care and the goal of personalized care. They saw that it created efficiencies by allowing the providers to move more quickly through visits when helpful family members were present. But they also described experiences when, although necessary, it resulted in inefficient care and produced chaotic interactions with the patient. Thus, based on their experience trying to involve family members in care, the staff saw this best practice as high risk.

Well, sometimes family members work out great; sometimes they don't. I have a lot of caregivers, and this is the one part that never works out great. It's when they bring in, usually it's daughters bringing in their mom, and they drop them off here and then they're going to get their grocery shopping done while the visit goes on. And you see the patient and go, “Honey, are you taking your medications every day?” “Yes.” “Good. Are you taking such and such?” And they go, “I don't know.” “Well, what do you mean?” Well, they don't know because her daughter or granddaughter ended up pouring out her meds. And I go, “OK. So you don't know exactly what you are taking?” “No.” That's an issue. Because then we've got to talk to a lot of those people who have direct involvement in that patient's care and tell them they have to be there in the office when we're discussing this. And it turns out the daughter, the granddaughter, everyone is involved in the care. It's a nightmare. And then when some family is in the office with the patient, I have to talk to each person, and ask, “Is she taking this med or that med?” And they'll say, “I don't know.” “What do you mean?” “Well, that's Mom's job to give to her at night: I'm there during the day.” Great, so then the granddaughter gets her mother on the phone, and the mother has her two great-grandsons in the room that are absolutely causing complete havoc. And this discussion is happening right in the exam room with the patient sitting there! It takes forever. (physician, practice 1)

In particular, the best practice became very inefficient and a “time sink” for staff when (1) too many of an elderly patient's family members were involved in their care and (2) family members were involved who conflicted or disagreed with their elderly loved ones or who should have been involved but were not willing to. In these situations, staff felt that using family members was more trouble than it was worth. They believed that too many of these situations on one day ruined their work flow and scheduling. As a result, each practice also thought that EMR information containing informal details about a patient's family was as important to use when deciding when not to involve them as when to involve them.

Finally, each of the practices to a lesser or greater extent felt that patient-centered care worked best with the elderly when it also used communication approaches tailored to an older patient's preferences. Because knowing these preferences was key to identifying the appropriate communication strategy, intimate knowledge of a patient also helped in the success of this best practice. When figuring out how to personalize their communication with elderly patients, the staff—especially physicians and nurses—relied on knowing the patients’ personality and how they wished to interact with the practice. The staff of all six medical home settings believed that there was not a “one-size-fits-all” approach to communicating with elderly patients. The experiences of most staff had convinced them that technology was of limited use for communicating with the majority of their older patients. Instead, they believed that electronic portals, email communications, and scheduling through the computer did not make their practices more accessible to this population.

This conclusion appeared to dissuade the staff from using technology or these electronic innovations for their elderly patients, although this feeling varied. For example, in a medical home practice in an affluent suburban area, with higher numbers of educated older patients (e.g., retired doctors, lawyers, and teachers), the staff was more apt to use technology to enhance access, as they believed that more of their elderly patients were comfortable using email, smartphones, iPads, and other equipment.

But in medical homes in poorer urban and rural areas, the high concentration of less educated and sicker elderly persuaded the staff that traditional modes of communication (e.g., phone, letter, and face-to-face interactions) delivered care more effectively.

Dr. Y still likes to send his older patients actual letters in the mail, with instructions or test results or answers to their questions. He feels that they prefer to get this information that way, that they're used to it, and it takes him a lot of time, but he does it. And I think his elderly patients really do appreciate it. They don't use computers; most don't even have a computer in their homes. You can't get in touch with them unless you use the phone and are lucky enough to have them pick up. (nurse, practice 2)

Discussion and Conclusion

The results of this study showed that from the perspective of primary care practice staff implementing medical home care on a daily basis, there may be no easy, standardized approach to making such implementation work, particularly for special patient populations like the elderly. While the six practices were already level 2 or level 3 NCQA-recognized patient-centered medical homes at the time of this study—meaning that each had achieved a satisfactory level of medical home care implementation in the eyes of external stakeholders—they all were incorporating a richer variety of best practices that derived their validity from the staff's work experiences and interactions with elderly patients over time. These included both “hard” and “soft” practices that, at least for the staff I interviewed, worked in tandem to provide a more effective patient-centered care experience for older adults.

Among other things, this observation highlights the importance of socially constructed best practices—that is, the implementation dynamics derived from ongoing social exchanges between the staff and patient and among the staff themselves—to making health care workers believe that they could deliver effective medical home care to older adults. Even when asked directly about “approved” PCMH implementation activities such as the use of electronic medical records or the presence or absence of a particular work policy or process, staff downplayed the standard medical home elements against which they were measured as accredited medical homes. They acknowledged that they needed to comply with NCQA criteria, for example, to “look good” to payers and the outside world. They also believed that some of these criteria were necessary for good medical home care, as indicated in their identification of important hard best practices that were consistent with the NCQA template. In particular, they acknowledged that establishing specific protocols and processes around important aspects of care such as triage and medication reconciliation heightened the consistency by which the practice carried out these tasks, particularly for older adults.

But when speaking about their older adult patients, they highlighted other foundational practices that, to them, served as a launching platform for a larger array of activities, several of which were NCQA-approved processes and others that were not captured through the NCQA standardized assessment template, such as empathetic and compassionate care and tailored communication approaches. In their experience, the softer implementation practices provided a supportive, informed context for a variety of additional hard and soft practices over time. In the staff's view, without these foundational practices, reliable PCMH care for older adults would be undermined and, in some cases, impossible.

If indeed the staff employs a set of deeper, complex foundational implementation practices in order to produce a high-performing medical home, we should find effective ways to incentivize these settings to move beyond a pure fixation on standardized assessment templates and toward also identifying and cultivating these deeper dynamics. The soft practices in my study refer to the heavy relational components of care quality that were identified as critical to not just effective medical home care but also good primary care generally (Starfield, Shi, and Macinko 2005).

The danger is that while the current primary care business model does not fiscally reward either the relational practices to the degree necessary or the elements that nurture them (e.g., the ability to spend time with patients and get to know them) (see Hoff 2010), current PCMH payment approaches that involve capitated care management fees or bonus payments for specific quality or utilization targets may also prove an infertile ground for their growth. This leads to a more speculative observation: we must not repeat the mistakes with PCMH care reimbursement that have been made with primary care payment over time. This could doom the model to failure because greater reimbursement will be less geared toward rewarding the relational aspects of PCMH care, which tend to be time intensive. In addition, focusing too much on rewarding infrastructure, process, and redesign, for example, could also undermine the promotion of implementation practices that are needed for PCMH success yet require a specific work culture and heavy doses of regular social interaction between individual patients and staff to remain vibrant.

For example, there may be no substitute for paying practices to spend more time with their patients. Effective medical home care, as others have shown, would result in higher patient contact and higher costs for primary care, with the cost savings coming from other parts of the system, such as hospital care (Reid et al. 2010). This raises the question of just how many dollars good medical home care may save the system in the long run. Good primary care delivery is not cheap, nor is prevention aimed at individuals and populations. If providers and staff see softer relational implementation practices as resulting in effective medical home care, then we will need to keep in mind such practices have to be cultivated and maintained for every patient. This is different from current medical home effectiveness outcomes, such as preventing the overuse of hospital and ED care (i.e., from which the cost savings associated with medical home care often come), which end up applying to only a few of the practice's patients.

My study also suggests that PCMH implementation variation is necessary and desirable given the different types of patients that typify primary care delivery (Bitton et al. 2012; Hoff, Weller, and DePuccio 2012). Gray, Weng, and Holmboe (2012) made a similar point when they demonstrated the importance of assessing patients’ different experiences in predicting effective medical home transformations. This is not a small point. The industry's increasing homogenization of the term medical home through standardized measurements and a core focus on several key imperatives such as the use of information technology and specific care protocols belies the fact that each patient is unique, with different preferences, needs, expectations, and values, and that these attributes affect the success of a patient-centered care approach (Stange et al. 2010).

Stressing generalizability and comparability across the medical home landscape currently is understandable, given the newness of the PCMH model to many clinicians and staff and the need to link the provision of such care to financial incentives in a fair, transparent manner. But it is the providers and their staff who discover each day through their interactions with different types of patients what works best in making the medical home model work. This discovery takes place each day across a thousand primary care practices. We therefore can expect that medical homes may bear little resemblance to one another below the surface yet still perform highly effectively.

This is where a sensemaking perspective rooted in better understanding the everyday experiences of clinical and nonclinical staff has value. These staff possess a great deal of tacit knowledge, because of their work experience about how to best put different medical home elements into play, particularly for special populations like the elderly. This knowledge requires greater explication to transform the medical home model in a manner appropriate to the future. At the least, future research should attempt to develop more of these local stories of PCMH implementation across different care populations and in different organizational contexts. Perhaps the best practices identified in this study can also be applied to other patient populations. But we currently lack these rich, descriptive data for medical home evaluations, making it seem that we already know how primary care practices and their staff will get the medical home job done.

Methodologically, this observation supports the greater use of qualitative inquiry (see Bitton et al. 2012; Crabtree et al. 2011; Hoff 2010) in studying medical home implementation. The PCMH as a model of care sits within a larger context of different patients and complex organizational realities that determine how things are done. Whether or not a sensemaking view is adopted, an emphasis on collecting primary data across various practice settings would better illuminate the situational, contingent nature of PCMH success. Nonetheless, such empirical approaches remain the exception rather than the norm, because much medical home evaluative research draws on secondary data, which tend to control for rather than incorporate the surrounding milieu of medical home care (Hoff, Weller, and DePuccio 2012).

Study Limitations

This study has several limitations. As a qualitative study, its generalizability may be limited. Data collection was limited to the primary care practice staff working in six NCQA-accredited patient-centered medical homes, so care should be taken in extrapolating the findings presented here to other medical homes and practices. At present, a large number of medical home transformations are occurring nationally within primary care. Not all of these transformations look the same or have the same incentives moving them forward. The small sample was necessary, however, in order to use the qualitative approach taken, which was intended to gain a deeper examination of everyday PCMH implementation experiences at the practice and individual staff member levels and to let individual staff members articulate what they have learned from them.

A second limitation is in my sensemaking approach. Because this approach relies on the more subjective aspects associated with workers’ experiences and the interpretations of those experiences, it may have produced findings about PCMH implementation practices that are specific to the study group involved. Individuals may make sense of their everyday realities in different ways, and they also may have fewer everyday work experiences than others. As does the first limitation, this affects the study's generalizability. Finally, the study does not link the perceived best practices identified with patient care outcomes. Thus, even though staff regards these practices as important to effective implementation, it is unclear how much they affect the types of outcomes that medical home care theoretically should improve.

In conclusion, the results of this study suggest that effective PCMH implementation relies on a variety of best practices that are not limited to the basic, perhaps oversimplified, content of standardized medical home assessment templates. In particular, this study suggests that everyday PCMH implementation in “medical homes” consists partly of activities and approaches that are identified, cultivated, and reinforced through the “on the job” experiences of staff trying to deliver patient-centered care to particular patient cohorts, in this case, older adults. When comparing the best practices identified here with standard best practices of a body like the NCQA, we can say that the normative way in which the field currently regards medical home implementation is incomplete. Such incompleteness, however, is an opportunity rather than a shortcoming. For example, we might use the current results to help further delineate PCMH assessment tools such as those produced by the NCQA in areas like patient engagement or access that are more ambiguous for special populations like the elderly. When we allow ourselves to look at the everyday world of PCMH implementation as experienced, interpreted, and acted on by the individuals immersed in it, we find much more information about how to fulfill the promise of the medical home ideal.