Support for Contextual Control in Primary Care: A Qualitative Analysis

Abstract

Providing support for high-level cognitive performance is largely missing in many decision support designs. Most development in this area is structured to minimize attention, decrease the need for deeper processing and limit intense goal-directed cognitive processing. However, from a dual process perspective, both automatic and deliberative processes need to be supported. The purpose of this qualitative analysis is to explore complex cognitive processing. We used the Contextual Control Model to guide the analysis. Transcripts from 33 taped primary care visits across 4 locations in the VA were analyzed using iterative process of construct and thematic development. Five themes related to high-level cognitive processes were identified: 1) Joint Exchange and Patient Activation; 2) Planning and Proactive Problem Solving; 3) Script and heuristic processing; 4) Time perspectives and 5) Uncertainty management. Results are discussed in terms of the need to support integrated views for complex situation mental models.

INTRODUCTION

Medical decision-making ranges from simple categorization and well-learned automated routines to complex, intensive deliberation. In the former sense, cognitive load is minimal and decision support should minimize attentional demands and increase automaticity. In the latter case, decision support must take the form of augmenting human thought, promoting deep processing, capturing attention, enhancing performance and designing for complexity. Learning the difference between the two is a significant issue for developers. Significant strides have been made in providing interfaces and functionality that supports reminding, linking information to action and ease of input of data. However, little work has been done on understanding the functional requirements for complex mental processing, such as pattern recognition in situations with significant variability, projection and simulation, planning and problem-solving.

Several lines of evidence are converging that suggest significant limitations on the degree to which the electronic medical record (EMR) can live up to expectations to improve care. First, several literature reviews on the topic of computerized decision support and patient outcomes have failed to find generalizable effects.^1,2 Second, EMR’s may actually cause harm. Several publications have noted the presence of unanticipated negative consequences of CPOE,³ higher mortality rates after implementation of a CPOE system,⁹ and reports of increased medication errors.⁴ Even more recently, multiple authors are noting increasing dissatisfaction with EMR’s for quality and safety reasons.⁵

One reason for limited impact on the distant end-points, such as mortality, adverse drug events, and outcomes is the failure of the system to support the most cognitive complex activities involved in healthcare delivery, including coordination, adaptive control, and planning.^6,7 Ironically, the EMR’s superior ability to capture large amounts of information as compared to a paper chart makes them harder to use because it does not effectively integrate the huge volumes of content simultaneously across time, information categories and functional relations. As a result, using these systems takes significant cognitive resources in real life, despite greater access. A provider has to keep in mind multiple clinical, communicative and quantitative tasks (e.g. decision-making, patient education, coordination and documentation) while trying to identify relevant information amongst a large pool of non-relevant and dispersed content. Most systems do not support this level of synthesis efficiently.

The purpose of this study is to use a theoretical perspective to identify the characteristics of complex performance in the patient-provider encounter. Specifically, we used the Contextual Control Model (COCOM) to characterize performance in primary care visits.^8,9

Contextual Control Model (COCOM)

The clinical healthcare environment is a “Joint Cognitive System” that includes the provider, electronic health record, auxiliary staff, work processes, and other infrastructure. Together, these components of the JCS function to maintain control and support performance toward accomplishing clinical goals. The electronic health record (EHR) is a key component of the JCS, providing patient-specific clinical information, serving as a platform for communication, and a central repository for recording clinical events. From the JCS perspective, the CPOE system is an active partner in the process of care. The provider adapts his or her behavior to the limitations of the CPOE system and also alters the functions of the information system to better meet their own needs. These reciprocal adaptations and work-arounds are often effective in improving the performance of the JCS, but also may come at a performance cost. ** (cost/benefit)

In the CoCOmControl is characterized as having 4 control modes in increasing order. In the scrambled mode (the lowest form of control), minimal information is available, goals or problems are underspecified and considered one at a time, time horizons are limited, and decision heuristics are rarely used. In the opportunistic mode, goals are poorly defined, only current information is considered, uncertainty is not well recognized, and decision heuristics are strongly influenced by habits and pattern recognition. In the tactical mode, goals are more defined, past and current information is considered, uncertainty is recognized and strategies are in place to cope with the uncertainty, but decision heuristics are rigidly based on guidelines. In the strategic mode, interactions among high-level goals are considered, timelines are much longer with significant planning and forecasting, and decision heuristics are flexibly adapted to the current context.² Movement between the modes is a function of the quality of planning, which, in turn is a function of the decisions made, the time available and the inherent goals.

Dual Process Model

Dual Process Cognitive models are a group of theories that integrate a wide range of psychological findings from the last 50 years in the areas of decision-making, self-regulation and behavior change.¹⁰ This group of theories posit the presence of two systems of reasoning or memory. The first arises from the vast set of associations and knowledge links that constitute memory or knowledge accumulated over a lifetime. Activation of these linkages (external or internal cause) creates a “spreading activation” across related concepts. The result is very fast pattern recognition, quick automatic behavioral responses (often without awareness) and a sense of intuitive “knowing.” This system is called “System 1” and is largely responsible for both expert knowledge (rapid pattern-matching) and as well as the common heuristics and biases (e.g. vividness). The other system is rule-based, conscious, and slow (System 2). System 2 comes into play when accomplishing goals that require more attention, deliberative reasoning and active learning. Learning varies across the two systems. In System 1, learning is a slow process of association and is often unconscious (multiple, repeated exposures). Learning in System 2 is a process of thinking and analyzing requiring significant conscious resources and effort, but can be very fast. These systems are continuous, simultaneous processes with humans generally motivated to avoid System 2 thought, if possible, in order to preserve cognitive resources. (Evans, 2008; Smith, et al 2000) Our general hypothesis is that decision support is important for both automatic (System 1) and deliberative (System 2) processes. In much of informatics, System 2 cognitive support is neglected.

METHODS

Design

This study was part of a larger study focusing on designing and developing an effective system of cognitive support for the electronic health record. For this portion of the study, a descriptive, observational design was employed. This approach is the most appropriate to efficiently gather information about provider’s mental models regarding medication management, goals associated with ordering and information search behavior, and information needs for the generation of hypotheses.

Settings

VA medical centers were chosen randomly based on their geographical distribution, size, and academic affiliation, i.e. they had resident training programs. Potential site PI’s were identified and asked whether or not they would like to volunteer to participate. Sites were identified and recruited until we obtained the participation of four sites that were somewhat evenly distributed according to size, location and presence or absence of resident training. In addition to the main site, Salt Lake City UT, the participating sites were: Asheville NC, West Haven CT, Seattle WA, and American Lake WA.

VA Electronic Medical Record

The VA’s Computerized Medical Record System (CPRS) is an integrated system, covering both inpatient and outpatient clinical areas. CPRS includes computerized provider order entry (CPOE), electronic documentation, consults, labs and integrated reports. Progress notes, procedure results could be printed, but are usually not used. Multiple levels of decision support are provided in CPRS, including drug-drug alerts and guideline decision support.

Participants

Within each site, up to 10 provider participants were selected based on their staffing in primary care outpatient clinics. Eight to ten primary care providers were recruited at each site: West Haven (N=9), Asheville (N=9), Salt Lake City (N=9), American Lake N=8, Seattle N=10), for a total of 45 providers. Provider participants were recruited by email and word of mouth. Employees’ supervisors were not asked to recruit participants. Additional provider participant selection criteria included being a prescribing provider and having at least one year of involvement in the VA. Patient involvement was based on provider participation and the presence of hypertension. Once a provider volunteered, one of his/her patients who met the criterion was approached and asked if he/she would like to participate in the study and consented.

A description of participating providers included in this analysis is presented in Table 1. Clinical pharmacists and a nurse case manager were eliminated because the focus was on the primary care visit itself and how the provider treated the disease(s) in question. In addition, there were several cases where the tablet data was not complete because of researcher error, leaving 33 transcripts to be included in this study. This study reports on the review of 10 transcripts.

Table 1.

Description of participants

Site	Attending	Residents	Physician Assistant	Nurse Practitioner
1	3		2	2
2	6		1	1
3	3	1
4	5			3
5	5			1

Data Collection Procedures

Observations

Primary care provider participants and their patients were observed and audio recorded during the visit. The research assistant was in the room and occasionally would ask clarifying questions post visit, but otherwise was silent. Only one observer attended any one patient visit. The research observer situated herself silently in the room and started the tape recorder and the PC tablet data collection program (in Access). Microsoft Access was used to build a data entry interface to code provider’s behavior and make notes about their observations in real time. Every provider had a computer in the patient exam room. The researcher could select when the provider was talking to or examining the patient and if the researcher left the room to give the patient and provider privacy. Qualitative notes and comments regarding the visit were also recorded on the tablet in narrative form.

Qualitative Data Analysis

The audio-tapes were transcribed and all identifying information removed. Four investigators used a qualitative software, ATLAS@ (http://www.atlasti.com), to review the transcripts independently at first and then again together to discuss and refine pre-code categories. The initial focus was on the 4 areas of COCOM areas of goal integration, time horizon, uncertainty regulation and decision processes.

However, analysis was not restricted to the four COCOM categories with the overall focus on identifying performance characteristics associated with complexity. Over 80 pre-codes were generated, distilled after discussion to about 24. Final discussion and review resulted in the following findings.

RESULTS

The final organization of themes is listed in Table 2 below. These themes identify broad categories of performance related to “control” in the primary care visit. High levels of control are associated with cognitive complexity at times and at other times are more automatic.

Table 2.

Identified Themes and related COCOM Constructs

Themes	COCOM Constructs
Joint Exchange and Patient Activation	Goal Integration, Time Horizon and Decision Processes
Planning and Proactive Problem Solving	Goal Integration, Time Horizon and Decision Processes
Script and Heuristic Processing	Decision Processes
Time Perspectives	Goal Integration, Time Horizon
Uncertainty Management	Decision Processes

Joint Exchange

Visit dialogues that were identified as being a “joint exchange” had the character of a shared or equal discussion between the provider and the patient. These topics could be anything and ranged from the patient telling their story and the provider prompting and clarifying, to decisions about treatment that were a “back and forth” of information exchange about what would work and would not work. For example, if they included an educational component, the interchange might be initiated by a question from the patient, clarification of the question by the provider, assessment of the patient’s knowledge, medical information given and further questions by the patient. Another category of conversation that was included in the category of Joint Exchange was the interactive summary. In some visits, the summary and/or plans appeared to be negotiated rather than just listed at the end. These types of exchanges are clearly different than what we called “MD directed” which was a set of directive questions requiring yes/no or limited responses. In the exchange below, the patient raised a topic and the provider and the patient discussed it. What we don’t have room to present is how the topic is readdressed at the end of the visit when another solution is offered by the provider and the patient decides against that one as well. They finally agree on a third option.

PATIENT: The other concerns I have...and that is...that is...that’s the head. It just seems...I mean I wear a hat. I definitely protect myself from the sun. I don’t get out in the sun that much anymore, right now anyhow. That’s...not that much more right now anyhow. Anyway, do the paperwork and everything else that I’m doing but it always...seems like... seems like I’ve got scabs constantly. And I mean it’s not coming from shaving. I mean, I shave but that’s not what’s causing it because otherwise I’d be having a bleeding effect. It’s just, uh...

PA3: Kind of scaly? Can you feel them more than you can see them?

PATIENT: Uh yep. Um-hmm, I’m starting to feel a few.

PA3: Okay.

PATIENT: And uh, that, that’s probably, that’s probably a...I gotta face the moment of truth, let you freeze them off I guess.

PA3: All right. You’ve had them frozen off before?

PATIENT: I hate it, yeah.

PA3: You know it blisters up, stings a little bit.

PATIENT: Yeah, it stings a lot. Stings a lot a bit for whatever reason.

PA3: It does, especially on your head...it’s pretty sensitive. Do you make sure you’re wearing a hat all the time? There’s not a lot up here to help protect your head and make sure you’re wearing some kind of sunscreen, long sleeve, long pants if you’re going to be out in the sun for a long period of time.

PATIENT: Yeah.

PA3: Okay I can take a look at those and freeze them off for you.

PATIENT: I’m not sure I want it done today.

PA3: Okay it’s up to you. If you don’t want to do them today, that’s okay too.

The joint exchange is a fairly high level conversation that includes patient education, assessment of patient preferences, patient activation and involvement. It usually involves a discussion of not just what the patient was doing or what clinical options are available, but also the whole health care delivery system. The wide range of information needed and the degree of integration required illustrates the complexity of thought. Besides the patient’s history and clinical condition, the provider needed to know how to access specialists, which clinics are located where, patient preferences and social situation, and what other team members are available to be enlisted for help. The “feel” was of densely interactive collaborative interchange that could be operationalized as the integration of the patient’s perspective, the clinical status, and the healthcare delivery situation using both temporal and operational views.

Planning and Problem-Solving

The planning and problem-solving category included a wide-range of conversation topics. Some problem-solving conversations involved diagnostic reasoning and problem identification whereas others involved dialogues about how to get things done in a timely or effective manner. The example presented below is typical of the planning and organizing dialogue.

PATIENT: Okay, there’s...uh, I believe they would like to schedule a colonoscopy, in the not-too-distant future. And uh, one of the medications that I’m taking, I have to be off of it.

PA3: That’s the Plavix.

PATIENT: It is the Plavix? Okay. Is that the thinner?

PA3: Um-hmm (yes).

PATIENT: Okay. Okay.

PA3: So what I would do, get an appointment with Cardiology and see what they say...

PATIENT: In terms...and then go from there?

PA3: Yeah, depending on what their recommendations are. If they feel it’s okay for you to stop Plavix...you’ll have to stop it 7 days before you have your procedure done and then usually you can restart it, but if they’re going to plan on stopping the Plavix altogether, then that’s a different story and you can have that done.

PATIENT: Why would they...why would they want to stop that?

PA3: The Plavix while they’re doing the procedure?

PATIENT: No. No.

PA3: Oh as far as..

PATIENT: Eventually taking me off of it. Yeah.

PA3: So usually people are on that after they have a stent placed for about a year afterwards. Depends on where they’ve put stents in different parts of your heart, and what they would recommend as far as continued follow-up if they wanted you to stay on it longer than a year so...

PATIENT: Okay.

PA3: Yeah some people are on it for longer and some people only take it for a year; it just depends on where the stents were placed.

PATIENT: Right. Okay.

PA3: So find out from them what their recommendations are. If they feel like it’s fine to stop it, let me know and you call the Tip Line and leave me a message and I can order that colonoscopy for you. If it’s something you’re going to stay on longer than that, then we’ll have to talk about what to do as far later enriching therapy and stuff.

The complexity in this interchange comes from the patient’s need for education, the primary care provider’s monitoring of subspecialists and the coordination regarding communication procedures. Note in this case, that a great deal of coordination is passed onto the patient, who is acting as an active participant. The provider may not be aware of how and who to utilize in terms of other members of the team, or institutional resources that could enhance communication processes.

Script and Heuristic Information Processing

After multiple reviews of the transcript, common “scripts” were relatively easy to detect. Scripts are highly patterned behaviors and structured interchanges. The common scripts that we identified were: 1) medication review; 2) prevention screening (depression, safety, etc.) and 3) review of systems. These interchanges were characterized by rather lower levels of complexity and had a “routine” feel to them. The example below was typical.

MD1: Are you still taking the Tylenol regularly?

PATIENT: Right.

MD1: Still taking this one for restless legs?

PATIENT: Right.

MD1: Okay. Hydrocortisone cream, do you still need that?

PATIENT: Yes.

MD1: Okay, well, we should renew that then because it’s expired.

PATIENT: Okay.

MD1: Okay, all right. What are you using that for?

PATIENT: Whenever I get a rash and stuff and it works on it(?).

MD1: Do you still take this calcium here?

PATIENT: Calcium chloride.

MD1: Are you taking the calcium tablet?

PATIENT: That last...thiazide.

MD1: Yeah, do you take that?

PATIENT: Yes.

MD1: Okay, and then you take an aspirin a day and a vitamin.

PATIENT: Aspirin a day...right.

MD1: All right, I’ll see if there’s anything else.

PATIENT: And the fish oil.

Time Perspectives

This category referred to dialogue that referenced time. Changing temporal perspective is a cognitively demanding activity and it is often required in order to get full situation awareness. Time perspective includes situations where providers would take an “if-then” view or would try to prognosticate about the likelihood of certain outcomes. In the visit room, patients are often asking for such projections and providers rarely have adequate information to answer the questions. Needed information would include population based data as well as knowledge about the trajectory of disease and the relative risk of side effects. Time perspectives also come into play with the relatively constant desire to understand change in symptoms and disease over time. The EHR does not easily provide that view, but often dialogue implied or directly referenced trends. Especially important are trends correlated with treatment changes, a view that is very rarely available. The result is providers engaging in the cognitive complex task of simulating that information themselves.

MD: Thyroid was normal and anemia is still a little bit anemic, but your...

PATIENT: I’ve had that over a period of years though.

MD: Yes, the numbers, the number is getting stable and in fact, the last one a couple of weeks ago was fine. It was normal.

----------------------------------------------------------------------------

PATIENT: Well actually, I started feeling maybe about a week or so. But more so today, because we have stairs at work, and going up those stairs, I may go up some stairs and stop. Catch my breath and then walk the rest of the stairs. And then, on my way to my locker, I might have to kind of pause a minute and go to my locker. So it’s going up, and then when you take breaks, you go up and down stairs. I hate _______. I’d give anything if I’d weigh 175.

PGY2: Yeah, me too. It looks like I’m just looking at this note from February, so you were admitted in clinic last February, right?

PATIENT: Yeah.

PGY2: How do you feel compared to then? Same, worse? Can you remember back to that day?

PATIENT: About the same. I mean...

PGY2: You’re not at your baseline though. You’re not at your baseline? Feel like your legs are more swollen?

PATIENT: Well, if you take those deals off, then it would be like a balloon. This time I’m too big.

Uncertainty Management

Identifying areas of uncertainty and addressing it is a high level of cognitive processing. It requires meta-awareness and enough attentional resources that can monitor one’s own failure to understand. Uncertainty can arise from incongruent or inconsistent data, lack of knowledge, desire for scientific information resources or simply lack of the necessary information. Uncertainty is an important component of effective decision-making in that we need to pay attention to the reliability and validity of the data.

PATIENT: By the way, the Plavix, for some reason I only got a 30-day supply. I normally get a 90-day supply.

PA3: Yeah, the formulary has changed on that one. Only do 30-day supplies, continue out for a year.

PATIENT: But it’s an automatic, right? Is it going to be automatically sent to me again?

PA3: I don’t know if they automatically will send that to you. Do you usually call in every month?

PATIENT: I called in this last time, and received all three medications with the exception of the Plavix.

PA3: Okay. I think you’ll probably still have to call in for that one.

-----------------------------------------------------------------

PA: Yeah. I just started feeling the changes, the difference in...I’m gaining weight, too. Before I was...

MD1: That’s supposed to be the other way, so then you don’t have enough thyroid you gain weight. When you have enough, you lose...well, then you don’t gain weight.

PA: Well, something’s up there.

MD1: But if you...maybe you’re gaining weight because you’re feeling better.

PA: And I’m eating more, too.

MD1: And you’re eating more.

PA: Well, I feel like I gained weight.

MD1: Yeah? In your belly you feel that or is it in your legs? Do you get swelling in your legs?

PA: Yeah, my leg swells up, too.

MD1: Okay, you know what, I think we can _______

DISCUSSION

Review of Findings

Patterns of interchange associated with high levels of control in a visit (strategic level) appeared to be associated with significant cognitive complexity. Some constructs, such as time perspectives were similar to the COCOM model, but others were broader, such as the joint exchange. However, even the later were congruent with the concepts of the Joint Cognitive Systems on which COCOM is based.^8,9,11

Implications for Cognitive Support

As Stead and Lynn noted in a paper reviewing the efficacy of Electronic Medical Records: “IT applications appear designed largely to automate tasks or business processes. They are often designed in ways that simply mimic existing paper-based forms and provide little support for the cognitive tasks of clinicians or the workflow of the people who must actually use the system.” (p. 3).⁶

Information integration in the user experience should make it easier to understand relationships among data to get a sense of patient status, identify goals, and plan interventions. Information integration is far more than presenting the right data in the right place at the right time. This typical mantra implies data access, juxtaposition, and filtering. These activities are just the beginning. The user must ingest the data and construct a mental model of what is going on with the patient. As the above themes make clear, the clinician not only must understand the medical model of the patient, but also the social model. The social model includes a sense of patient education needs, goals, preferences, and constraints on following medical advice. Of course, data in this form are rarely found in typical EMRs and never in an integrated view.^12–14

Information integration must also adapt to user needs to support the appropriate point on the spectrum of heuristic versus deliberative processing. In the case of scripts, information presentation should make it very fast for the user to orient to patterns of information then select and complete tasks. When dealing with uncertainty, however, information needs to be organized to facilitate exploration and deliberation. The EMR can help the user walk through hypothesis generation and testing by bringing into consideration sets of related information that support or disconfirm ideas. Likewise, when the user appears to be moving in the wrong direction, the user experience could present more disconfirming information to force more deliberation.

Implications for Informatics and EHR Design

These findings have two significant implications for the Informatics community and for design of the electronic medical record. First, designing static displays for every clinical situation is not feasible and will likely not be effective. Providing multiple layers of attributes and links will help the user identify where information is so searching for what seems like disparate information is minimized (e.g. finding services for homeless veterans when setting up procedures for a colonoscopy). Secondly, users need substantial more tools for manipulating the information environment than they currently have access to. Setting up processes for self-reminders, tools for organizing information easily into one view, providing support for simulation or “what-if” questions are only a few possibilities of this category.

Limitations

The sample size in this study was small, although geographically diverse. The focus was on the visit dialogue and not on many other aspects of the care process, such as preparation, data analysis, team planning and other care processes.

CONCLUSIONS

Future work on designing decision support should focus on balancing both System 1 and System 2 information needs. Information integration, or framing at high levels of complexity are necessary for many of the activities in healthcare processes.