SEMIOTICS OF ‘NONCOMPLIANT’ PATIENT

Charulata Ramaprasad

University of Chicago, 5841 S. Maryland Avenue, Chicago, IL 60637, U.S.A.

Arkalgud Ramaprasad

University of Illinois at Chicago, 601 S Morgan Street (MC 294), Chicago, IL 60607, U.S.A.

Keywords: Noncompliance, Semiotics, Medical Errors, Electronic Medical Records.

Abstract: ‘Noncompliant’ (NC) patient is a common label in medical records. While it encapsulates many dimensions

of undesired patient behavior, the semiotics by which it is generated and applied is unclear: What data

indicate noncompliance? How are the data analyzed and interpreted to label a patient as noncompliant?

How does the label frame the physician’s thinking? How does it affect the physician’s diagnosis, treatment,

instructions and actions? How does it affect medical outcomes? This lack of semiotic clarity can result in

medical errors. We provide a framework (a) for conceptualizing the semiotics of NC, and (b) to understand

the sources of potential medical errors. We illustrate the framework with a case study. The framework can

be used to manage noncompliance effectively and reduce medical errors, especially with EMRs.

1 INTRODUCTION

‘Noncompliant’ (NC) patient is a common label in

medical records. Noncompliance, from which the

NC label is derived, is of major concern in medical

care. It “is a value-laden term, heavily weighted

against the patient who, by definition, refuses to

yield or conform to doctor’s advice.” (Hill, 2004, p.

2004) It can diminish the effectiveness of the most

efficacious course of treatment, and the efficiency of

medical care delivery (Cleemput & Kesteloot, 2002;

Reach, 2008). It is a pervasive problem affecting

almost all types of care and populations (Rosner,

2006). This paper explores the translation (a) of

noncompliant behavior to the NC label in a medical

record, and (b) of the effect of NC label in medical

practice.

Labeling a patient NC almost invariably implies

that the patient does not have a good reason for

his/her behavior – that the behavior is irrational and

difficult to manage. In contrast, recognition that a

patient has been noncompliant occasionally but for

‘good reason’ – he stopped his anti-hypertensives

because he lost his insurance coverage for

medications, she does not exercise because her

neighborhood is unsafe, he does not follow

instructions for his colonoscopy because he is

illiterate – allows one to attempt to fix the problem.

Noncompliance encapsulates many dimensions of

patient behavior which can be summarized in a

simple ontology (Figure 1) constructed from the

literature. The ontology has six dimensions of

noncompliance: (a) Visits, (b) Medications, (c)

Tests, (d) Procedures, (e) Lifestyle, and (f)

Administration. Each dimension has a list of

possible categories of noncompliance. In visits, for

example, a patient may be noncompliant in one or

more of the component categories: preparation for

the visit, interaction during the visit, follow-up

actions, and scheduling. A patient’s noncompliance

profile would be the aggregate of his or her

noncompliance on the categories within all the

dimensions. The aggregation is indicated by the [+]

sign between the columns in the ontology.

Patients may have different noncompliance profiles.

Some may be simply noncompliant in taking the

right medication dose. Some, on the other hand, may

be noncompliant on many categories of many

dimensions. Thus, a patient may be noncompliant in

visit preparation, medication refills, test follow up,

and diet lifestyle. The ontology encapsulates a very

large number of possible profiles. It is likely that

only some will result in the patient being labeled

NC. This paper addresses the process by which the

noncompliance profiles are translated into the NC

label for a patient. Patient noncompliance has to be

257

Ramaprasad C. and Ramaprasad A. (2010).

SEMIOTICS OF ‘NONCOMPLIANT’ PATIENT.

In Proceedings of the Third International Conference on Health Informatics, pages 257-262

DOI: 10.5220/0002734902570262

 SciTePress

Visits Medications Te sts Procedures Lifestyle Administration Explanatory Factors

Preparation [+] Administration [+] Preparation [+] Preparation [+] Diet [+] Payments Patient factors

Interaction Dosage Conduct Interaction Exercise Paper work Treatment factors

Follow up Refill Follow up Follow up Tobacco Lifestyle factors

Schedule Schedule Schedule Schedule Alcohol Demographic factors

Monitoring Sociodemographic factors

Psychosocial factors

[due to]

[Noncompliance in]

Noncompliance Dimensions

Figure 1: Ontology of Patient Noncompliance.

understood in the context in which it occurs.

Sometimes there may be good reasons for a patient’s

noncompliance as mentioned earlier. The

explanatory factors for noncompliance are listed as a

separate dimension with six categories based on

Rosner (2006). They are (a) Patient factors, (b)

Treatment factors, (c) Lifestyle factors, (d)

Demographic factors, (e) Sociodemographic factors,

and (f) Psychosocial factors. This dimension is

connected to the others by the phrase ‘due to’, as

shown in Figure 1. A detailed list of the

subcategories of explanatory factors is given by

Rosner (2006).

The ontology can be extended and refined by

modifying the dimensions and the categories within

them. Some dimensions in the ontology may not fit

some contexts – the Procedures dimension may not

fit a non-surgical context. Or, a specific subcategory

of explanation such as lack of transportation may be

required for rural patients. Once the ontology has

been adapted to a context it can be used to profile a

patient’s noncompliance and the factors which

explain it. It can also be used to modify physician

and patient behavior in a way that satisfies both

parties.

We will use the ontology as a framework for

discussing the semiotics of labeling a patient NC.

Pragmatics

(Framing

presuming

noncompliance)

Syntactics

(Relationships

indicating

noncompliance)

Pragmatics

(Labeling patient as

noncompliant (NC))

Morphologics

(Data indicating

noncompliance)

Semantics

(Interpretation as

noncompliance)

Patient characteristics

Observations

Actions

Semantics

(Diagnosis

presuming

noncompliance)

Syntactics

(Treatment plan

presuming

noncompliance)

Morphologics

(Instructions

Presuming

noncompliance)

Pragmatics

(Framing

presuming

noncompliance)

Syntactics

(Relationships

indicating

noncompliance)

Pragmatics

(Labeling patient as

noncompliant (NC))

Morphologics

(Data indicating

noncompliance)

Semantics

(Interpretation as

noncompliance)

Patient characteristics

Observations

Actions

Semantics

(Diagnosis

presuming

noncompliance)

Syntactics

(Treatment plan

presuming

noncompliance)

Morphologics

(Instructions

Presuming

noncompliance)

Figure 2: Semiotics of ‘Noncompliant’ Patient.

The paper is focused on two questions: How is a

patient labeled NC? How can the label affect his or

her medical care and hence medical outcome?

2 SEMIOTICS OF

‘NONCOMPLIANT’ PATIENT

Semiotics has a long history in medicine (Hess,

1998). It is the study of how information is

generated about objects and applied to actions on

them. This paper focuses on how noncompliance

information is generated about patients and applied

to their medical care. Generation and application are

the two phases of the semiotic cycle as shown in

Figure 2. Each phase is sequential and has four steps

– morphologics, syntactics, semantics, and

pragmatics. The sequence of these steps in

application is the opposite of that in generation as

shown in Figure 2 (Ramaprasad & Ambrose, 1999;

Ramaprasad & Kashyap, 2008; Ramaprasad & Rai,

1996; Stamper, 1973).

In generating the label NC for a patient,

morphologics is the process of obtaining data

indicating noncompliance; syntactics is the process

of discovering relationships within the data

indicating noncompliance; semantics is the process

of interpreting the relationships as noncompliance;

and pragmatics is the process of labeling the patient

as NC. In application of patient’s NC label,

pragmatics is framing the problem presuming

noncompliance; semantics is the diagnosis

presuming noncompliance; syntactics is the

treatment plan presuming noncompliance; and

morphologics is the instructions presuming

noncompliance.

Consider, for example, a young male unemployed

patient who has missed a few follow up visits over

five years, has not refilled his medication

intermittently during an unknown period, continues

to smoke, and is delinquent on payments. Is he a NC

patient or simply a patient who has been

noncompliant occasionally?

One cannot answer the questions without

collecting data from a number of sources, in

different locations, on different media, in varying

HEALTHINF 2010 - International Conference on Health Informatics

258

formats, over a long period of time – corresponding

to the dimensions and categories in the ontology

(Figure 1). Even then, the data may neither be

complete nor accurate (Smith et al., 2005). It may be

difficult, time consuming and costly for a person to

systematically analyze the data to generate a profile,

interpret the profile, seek explanations, and make a

reasoned judgment whether he is NC or not. A

comprehensive EMR could reduce the time, effort,

and cost it takes to document reasons for

noncompliance and disseminate that information to

other members of the medical team to whom it is

relevant (social workers, for instance). In its

absence, the health care provider will likely make an

intuitive judgment based on limited data about

noncompliance behavior, its analysis, interpretation,

and possible explanations. Even with an EMR, if it

lacks the prompts necessary for an investigation into

the reasons that underlie noncompliance, the NC

label will be generated and recorded and is likely to

be applied without further validation or

investigation. It is no more ‘as if’ the patient is

noncompliant, the patient ‘is’ NC. The label is

attributed to the patient – the patient becomes NC

(Ramaprasad, 1987). Any nuances in the patient’s

noncompliance and possible explanations for such

behavior are also lost in the shadow of the NC label.

It creates a prospective expectation of

noncompliance. Further, the label is propagated in

the records without question or revalidation. EMRs

render the propagation easy and efficient with their

ability to ‘copy-and-paste’ (Hirschtick, 2006). The

incorrect labeling may be an unintended error, but

an error nonetheless (Ash, Berg, & Coiera, 2004).

Once labeled NC, future interactions with the

patient will likely be framed with the presumption of

general noncompliance, not necessarily of a specific

type or for a particular reason. As a consequence,

the diagnosis, treatment plan, and instructions will

likely presume noncompliance too. This

presumption can lead to medical errors in a number

of ways – less attention may be paid to the

diagnosis, treatment plan, and instructions due to

lower expectations of the desired outcome or a less

than optimal path of action may be chosen to

accommodate the prospective noncompliance. These

may be preventable rational errors (Federspil &

Vettor, 2008).

In the following we illustrate the semiotics of NC

patient with a case study conducted by the primary

author in a major urban hospital. Subsequently, we

analyze the case using the above framework and

suggest how similar errors can be avoided in the

future. In conclusion, we discuss how EMRs can

incorporate semiotics to avoid incorrect and

inappropriate NC labeling and its consequent errors.

3 CASE STUDY

A 19 year old male with a three-year history of

Crohn’s Disease (CD) was hospitalized in July 2005

with abdominal pain and pneumaturia (Figure 3,

Hospitalization #1). He had visited the emergency

room (ER) once and the outpatient clinic twice

before his hospitalization. In the intake history for

the hospitalization he was noted to be NC. The

patient was diagnosed with a fistula, treated and

released with specific follow-up appointments

scheduled and a documented treatment plan. He kept

the follow-up appointments with the pharmacy (1

visit) and the outpatient clinics (3 visits).

About one month later he was readmitted to the

hospital (Hospitalization #2) with back, flank and

abdominal pain. He was again labeled NC. Physical

examination revealed fever, high heart rate and low

blood pressure. Imaging of the abdomen

demonstrated multiple abscesses in the muscles of

the abdomen and back as well as in the kidney. The

hospital course was complicated by a staphylococcal

blood stream infection. Despite the findings, which

typically require weeks to months of intravenous

therapy, the patient was discharged with seven days

of oral antibiotics and no scheduled follow-up

appointments.

Two months later the patient was admitted for the

third time (Hospitalization #3) with abdominal pain

and frank bleeding from the gastrointestinal tract.

He was again labeled NC. Abdominal imaging

revealed enlargement of the previously noted

abscesses and multiple new abscesses. Surgical

intervention was discussed but due to a family

emergency the patient requested discharge and

readmission. Three days of antibiotics were

provided to the patient, as were readmission papers

for the following week. No follow-up appointments

were arranged.

Again two months later the patient was admitted for

the fourth time (Hospitalization #4) with severe

abdominal pain, nausea, vomiting, and inability to

walk. In the two months he had (a) come to the

ER once and left before being treated, (b) been

turned away by the admitting office, and (c) been

treated once as a urology outpatient. The previously

demonstrated abscesses were larger and several new

SEMIOTICS OF 'NONCOMPLIANT' PATIENT

259

Figure 3: Schematic Timeline.

ones were present. Surgery was planned, but the

patient expired suddenly prior to surgical

intervention. The cause of death was sepsis due to

persistent abdominal abscesses.

The patient’s social history had been recorded

during Hospitalization #3. He had no relationship

with his parents, lived with his sister 14 miles from

the hospital, and was the primary caregiver for his

niece. He had 7

grade education, was unemployed,

and had no insurance. He was noted as suffering

from depression but had not received psychiatric

care or medication.

4 ANALYSIS

Three critical questions arise in assessing the effect

of NC on this case. First, why was this patient

repeatedly labeled NC? Second, what factors

contributing to noncompliance could have been

addressed, had they been recognized? Third, what

impact did the NC label have on the care provided to

this patient?

Ascertaining the specific reason the patient was

labeled NC during his initial hospitalization for CD

is difficult. The patient was young and often

emotional, which may have contributed to the

labeling – but was there data to indicate

noncompliance?

Prior to the patient’s Hospitalization #1 he was

seen three times (in the ER or ambulatory care

setting). After the Hospitalization #1, he presented

to the pharmacy to fill prescriptions and to all of his

scheduled clinic appointments (circled in Figure 3).

After the Hospitalization #2 he did not follow-up in

the outpatient clinic. Recall, however, that no

follow-up appointments were provided. After

Hospitalization #3 (during which he was discharged

early due to a family emergency), he attempted to

get care in the ER but left due to a long wait; he

returned to admitting with his admission papers and

was turned away; he was seen in follow-up clinic

with abnormal vital signs but was sent home (circled

in Figure 3).

Why, given the above data, was this patient

persistently labeled NC? Follow-up care is only one

aspect of compliance. A strong suspicion of

medication noncompliance was noted in the

patient’s chart. Assuming this was the reason for

the NC label, what explanatory factors related to

noncompliance could have been addressed?

Could the demographic and lifestyle factors,

noted during Hospitalization #3, have contributed?

It is known that family structure leads to improved

medication compliance in adolescents (Mackner,

Crandall, & Szigethy, 2006). Furthermore, patients

with CD have a higher lifetime prevalence of

depression than those without CD and depression is

a risk factor for medication NC (DiMatteo, Lepper,

& Croghan, 2000; Kurina, Goldacre, Yeates, & Gill,

2001).

Did the label of NC affect the quality of care this

patient received? The most obvious ways in which

the NC label affected this patient’s care relate to the

suboptimal treatment plans developed for this

patient. Lack of appropriate instruction to this

patient is its corollary. Surveys administered to CD

patients indicate that almost all CD patients desire

more information about their disease (Jones,

Gallacher, Lobo, & Axon, 1993; Scholmerich,

Sedlak, Hoppeseyler, & Gerok, 1987). However age

and grade appropriate information was not provided

to this patient.

The semiotic process appears to have failed the

patient at almost every step:

• Generation morphologics – There is no evidence

of noncompliance data being collected

HEALTHINF 2010 - International Conference on Health Informatics

260

systematically prior to labeling the patient NC.

These data and potential explanatory factors

were available.

• Generation syntactics – There is no evidence of

the noncompliance data being systematically

analyzed to profile the patient’s noncompliance

and its relation to patient’s social history.

• Generation semantics – There is no evidence of

the interpretation of the profile logically as

indicating NC.

• Application pragmatics – The label NC was

applied and propagated without investigation.

There were enough data to raise doubts about the

patient’s NC label.

• Application semantics – It is doubtful that the

NC label affected the pre-existing diagnosis in

this case.

• Application syntactics – The shortcomings of the

treatment plan cannot be unequivocally

attributed to the NC label. It is a possibility.

• Application morphologics – The shortcomings of

the follow-up instructions cannot be

unequivocally attributed to the NC label. It is a

possibility.

In the concluding section we discuss briefly how the

semiotics of ‘Noncompliant’ patient can be managed

more effectively and efficiently.

5 CONCLUSIONS

The semiotics of NC is as important to analyzing the

case as the failure of the systems which supported it.

It was an avoidable error which instead of being

corrected by available accumulating evidence to the

contrary, persisted on paper and in the EMR. The

evidence, apparently, was not analyzed and

interpreted – the clinicians along the chain were

neither alerted nor possibly motivated to do so.

Noncompliance was not seen as a hypothesis to

be tested, among many other clinical hypotheses, at

each stage; it was seen as a conclusion. The label

NC became a reality instead of the noncompliant

behavior remaining simply a possibility with

explanation. While framing the patient as NC could

have affected his treatment and instructions the

effects of such framing are yet to be established

(McGettigan, Sly, O'connell, Hill, & Henry, 1999).

The permanence of a patient’s health information

is a key strength of the EMR – ideally it should be

available and accessible every time everywhere.

Information not only persists in these records for

ever but can also be propagated everywhere. For

‘good’ information these properties are extremely

desirable; for ‘bad’ information they can be

extremely dysfunctional. Unless the systems have

the cognitive (Patel & Bates, 2003) ability to

recognize and correct the errors, and expunge them

from everywhere the record has been propagated,

these errors can cumulate over a person’s lifetime.

This requires the system to be semiotically self-

reflective and thereby self-corrective, but also to be

available to the patients or their designees to focus

their cognitive faculties upon the problem to allow

them to question and correct it. They could provide

a profile of noncompliance instead of propagating

the NC label.

REFERENCES

Ash, J. S., Berg, M., & Coiera, E. (2004). Some

Unintended Consequences of Information Technology

in Health Care: The Nature of Patient Care

Information System-related Errors. J Am Med Inform

Assoc, 11(2), 104-112.

Cleemput, I., & Kesteloot, K. (2002). Economic

implications of non-compliance in health care. The

Lancet, 359(9324), 2129-2130.

DiMatteo, M. R., Lepper, H. S., & Croghan, T. W. (2000).

Depression is a risk factor for noncompliance with

medical treatment: meta-analysis of the effects of

anxiety and depression on patient adherence. Arch

Intern Med, 160(14), 2101-2107.

Federspil, G., & Vettor, R. (2008). Rational error in

internal medicine. Internal and Emergency Medicine,

3(1), 25-31.

Hess, V. (1998). Medical semiotics in the 18th century: A

theory of practice? Theoretical Medicine and

Bioethics, 19(3), 203-213.

Hill, J. (2004). Non-compliance. The Lancet, 363(9425),

2004-2004.

Hirschtick, R. E. (2006). Copy-and-Paste. JAMA, 295(20),

2335-2336.

Jones, S. C., Gallacher, B., Lobo, A. J., & Axon, A. T. R.

(1993). A Patient Knowledge Questionnaire in

Inflammatory Bowel-Disease. Journal of Clinical

Gastroenterology, 17(1), 21-24.

Kurina, L. M., Goldacre, M. J., Yeates, D., & Gill, L. E.

(2001). Depression and anxiety in people with

inflammatory bowel disease. J Epidemiol Community

Health, 55(10), 716-720.

Mackner, L. M., Crandall, W. V., & Szigethy, E. M.

(2006). Psychosocial functioning in pediatric

inflammatory bowel disease. Inflamm Bowel Dis,

12(3), 239-244.

McGettigan, P., Sly, K., O'connell, D., Hill, S., & Henry,

D. (1999). The Effects of Information Framing on the

Practices of Physicians. Journal of General Internal

Medicine, 14(10), 633-642.

SEMIOTICS OF 'NONCOMPLIANT' PATIENT

261

Patel, V. L., & Bates, D. W. (2003). Cognition and

measurement in patient safety research. Journal of

Biomedical Informatics, 36(1-2), 1-3.

Ramaprasad, A. (1987). Cognitive Process as a Basis for

MIS and DSS Design. Management Science, 33(2),

139-148.

Ramaprasad, A., & Ambrose, P. J. (1999). The Semiotics

of Knowledge Management. In Proceedings of the

Ninth Workshop on Information Technology and

Systems. Charlotte, North Carolina.

Ramaprasad, A., & Kashyap, V. (2008). Semiotic Web for

Translational Medicine. In J. Suermondt, R. S. Evans

& L. Ohno-Machado (Eds.), AMIA 2008 Annual

Symposium Proceedings (pp. 1103). Washington D.C.,

USA.

Ramaprasad, A., & Rai, A. (1996). Envisioning

Management of Information. Omega, International

Journal of Management Science, 24(2), 179-193.

Reach, G. (2008). Patient non-adherence and healthcare-

provider inertia are clinical myopia. Diabetes &

Metabolism, 34(4, Part 1), 382-385.

Rosner, F. (2006). Patient Noncompliance. The Mount

Sinai Journal of Medicine, 73(2), 553.

Scholmerich, J., Sedlak, P., Hoppeseyler, P., & Gerok, W.

(1987). The Information Needs and Fears of Patients

with Inflammatory Bowel-Disease. Hepato-

Gastroenterology, 34(4), 182-185.

Smith, P. C., Araya-Guerra, R., Bublitz, C., Parnes, B.,

Dickinson, L. M., Van Vorst, R., et al. (2005).

Missing Clinical Information During Primary Care

Visits. JAMA, 293(5), 565-571.

Stamper, R. (1973). Information in business and

administrative systems: John Wiley & Sons, Inc. New

York, NY, USA.

HEALTHINF 2010 - International Conference on Health Informatics

262