Problematic Resident-patient Relationships

The Patient's Perspective


  • Poster presented at the 22nd annual meeting of the Society of General Internal Medicine, San Francisco, April 29–May 1, 1999.

    Presented at the 5th annual National Research Service Awards Trainees Research Conference, Chicago, Ill, June 25, 26, 1999.

Address correspondence and reprint requests to Dr. Boutin-Foster: Weill Medical College of Cornell University, 525 East 68th St., Box 46, New York, NY 10021 (e-mail:


OBJECTIVES: The objectives of the study were to identify the characteristics of a problematic doctor-patient relationship from the perspective of primary care patients who are cared for by medical residents and to determine whether patients' perception of the relationship is a function of their demographic, clinical, or social attributes.

DESIGN: Cross-sectional survey.

SETTING: An adult primary care practice in an academic medical center.

PATIENTS: One hundred fifty-one patients whose primary care physicians were senior internal medicine residents.

MEASUREMENTS AND MAIN RESULTS: Patients completed a questionnaire addressing several aspects of their doctor-patient relationship, the general health perception item on the SF-12, and items on social support from the Duke Social Support and Stress Scale. By design of the study, approximately half of the patients had been identified by their physicians as being in problematic relationships (n = 74) and half as being in satisfying relationships (n = 77). Among patients in relationships described as satisfying by their resident, 10% viewed the relationship as problematic. Of the patients involved in relationships described as problematic by the resident, 23% viewed their relationship as problematic (P = .03). Patients who rated the relationship as problematic were much more likely to also report low social support compared to patients involved in relationships described as satisfying (76% vs 16%; P < .001). Compared to residents involved in relationships described as satisfying by their patients, residents in problematic relationships were more likely to be described as being less accessible and less capable of handling medical complaints (P <. 001).

CONCLUSIONS: Patients were more likely to describe the doctor-patient relationship as problematic if they felt that the resident was less accessible or less capable of handling medical complaints, or if they had low self-perceived social support.

The outpatient setting provides an opportunity for residents to develop their skills as primary care physicians and to develop long-term mutually satisfying relationships with their patients. Unfortunately, misunderstandings and miscommunication can lead to feelings of frustration and anxiety in the patient, the physician, or both parties. This turn of events may diminish any potential benefit of the relationship.1,2

Prior studies on problematic doctor-patient relationships have been conducted mainly from the perspective of the physician.3–6 Only a few studies have asked patients to give their assessment of a relationship described as difficult by their physician.4,7,8 However, these studies were conducted among attending physicians with less focus on difficult resident-patient relationships.

Perceived health status has been shown to influence patient behavior and attitudes regarding medical care. Studies looking at the association between other patient characteristics, such as ethnicity, age, and gender, and their view of their relationship have found inconsistent results.3,9–11 Perceived social support has been linked with health outcomes such as adherence to medication, coping, anxiety, and utilization of health care.12–18 Little has been done to examine the association between perceived social support and patients' views of their doctor-patient relationships.

Previously, we asked internal medicine residents to give their opinions on the relationships with patients that were satisfying and those that were problematic.19 This study builds upon our prior findings. While physician-experienced difficulties in the doctor-patient relationship have been well documented, 3 questions remain unanswered. First, what do the patients involved think about their relationship with their physician. Second, what are some potential reasons why patients may be dissatisfied with care. Third, what are the physician characteristics from the patient's point of view that may contribute to problems in the relationship. The current study was undertaken to address these questions. Specifically, the objective of this study was to identify the characteristics of a problematic doctor-patient relationship from the perspective of primary care patients who are cared for by medical residents and to determine whether patients' perception of the relationship is related to their demographic, clinical, or social characteristics.


Description of Practice

The study protocol was approved by the Institutional Review Board at our institution. The study was conducted in the adult primary care practice of the New York-Presbyterian Hospital, an urban academic teaching hospital. One hundred twenty-six residents rotate through this practice. Each resident sees his/her own panel of patients under the supervision of a precepting attending physician. When patients are assigned to a resident, the resident-attending preceptor team becomes responsible for their care for the remainder of the resident's training.

Assembly of Patient Population

Senior residents were asked to select 2 outpatients with whom they regarded their relationship to be most problematic and 2 patients with whom they regarded their relationship to be most satisfying. In an effort to ensure that the classification of patients would be based on established relationships rather than on single encounters, we limited the selection to patients with whom residents had had at least 3 visits. In total, 238 patients were identified (117 in problematic relationships and 121 in satisfying relationships).

Patients received an introductory letter inviting them to participate in a study designed to better understand characteristics that they found to be either problematic or satisfying in their relationship with their primary care doctor. Patients were unaware of how they had been classified by their physicians. Patients were approached at the time of their next visit, and those who consented completed a patient questionnaire.

Several mailings, followed by telephone calls, were made in an attempt to contact all patients. In addition, the National Death Index for patients was searched. As a result, 151 patients (63%) completed a questionnaire. Among the remaining 87 (37%), 3 were deceased, 10 refused, and 31 had moved without a forwarding address or telephone number, leaving 43 patients who could not be located.

Patient Questionnaire

Patients were asked to describe their relationship as either problematic or satisfying. Similar single-item scales have been used previously in evaluating doctor-patient relationships.18 Patients also completed a 35-item questionnaire to measure additional aspects of the relationship. Seventeen items asked about components of the doctor-patient relationship, including the physician's availability and interpersonal and technical skills. These variables were selected because they represent important dimensions of patient satisfaction and quality of care.11,20 To determine the association between patients' perception of their social support and their perception of their doctor-patient relationship, 12 items on social support from the Duke Social Support and Stress Scale were asked.21 This is a reliable and valid scale that measures patients' perceived emotional support. The social support component consists of 12 items with 3 response options (none, some, a lot) and is scored from 0 to 100, higher scores indicating greater social support. For the purposes of this study, we were interested in perceived emotional support, which is defined as the availability of people an individual trusts, on whom he can rely, and who make him feel cared for and esteemed as a person.13 To determine the association between perceived health and patients' perception of the relationship, 1 item from the SF-12 that rates general health from excellent to poor was asked.22 The remaining 5 items focused on patient demographic characteristics.

The elapsed time between patient identification and completion of the patient questionnaire spanned from 1 to 4 months. If patients did not have a scheduled visit within this time period, they were contacted by telephone. To ensure that the patients were commenting on the same relationship identified by the resident, we asked them if there was one person they considered to be their primary care physician and to name that individual. If they responded with the names of more than 1 physician, they were given the name of the pertinent resident.

Using our computerized data management system (CLIMACS @ J. Hollenberg, Roslyn, NY), we determined the total number of visits made to the practice over the past 2 years and the most common diagnoses made during those visits. From this database, we extracted International Classification of Diseases, Ninth Revision (ICD-9) diagnoses, and using ICD-9 codes, we calculated a Charlson comorbidity score, as previously done by Deyo et al.23,24

Statistical Analysis

The student's t test was used for the comparison of continuous variables between patients who rated their doctor-patient relationship as problematic and patients who rated their doctor-patient relationship as satisfying. χ2 tests were used to compare categorical variables. Logistic regression analyses were used to identify predictors of a problematic relationship from the patient's perspective. In all models, the dependent variable was the patient's perception of the relationship, and the independent variables included patients' demographic and clinical characteristics, as well their perception of their health and social support. In these analyses, we also included as predictor variables physician characteristics such as age, gender, ethnicity, and whether they were primary care or categorical residents.

To determine which independent variables could predict how a patient perceived his/her doctor-patient relationship, 3 steps were followed. First, potential independent variables were screened as individual predictors in bivariate logistic regression models. Second, the predictor variables that met the threshold of P < .01 on bivariate analysis were examined for multicollinearity among themselves using Spearman's rank correlation. All remaining variables from this step that were not significantly correlated were evaluated in a multiple logistic regression model. Backward elimination was used to drop variables, starting with the least significant. Variables that remained significant at a level of P = .05 on final multivariate analysis were retained as significant predictors of problematic relationships from the patient's perspective. All regression analyses were done using the PROC GENMOD in SAS (SAS Institute, Cary, NC). Since each resident identified 4 patients, the REPEAT function in SAS was used to account for repeated observations.


Demographic Characteristics of Participating Patients and Residents

A total of 72 medical residents in their final year of training participated in the patient identification process. The mean age of the residents was 29 (SD = 2.3), and 44% were female. The majority were white (70%); 1% were African American, 6% were Latino American, and 23% were Asian American. Ten percent of the residents were in the primary care track.

A total of 151 patients completed questionnaires. The majority of patients were female, and the mean age was 56 years. Thirty-eight percent of patients were white, 29% were African American, 29% were Latino American, and 4% were Asian American. Patients were insured predominantly by Medicare or Medicaid, and 78% were unemployed due to disability or retirement. Most patients (89%) had completed high school. The most common reasons for visits included hypertension, back pain, depression, diabetes, and chest pain. The median Charlson comorbidity score was 0, with a range of 0 to 3. The most common comorbid conditions accounting for a score of 1 or more were diabetes, prostate cancer, and breast cancer. The average duration of the relationship between the patient and the identifying resident was 1.6 years (SD = .9). The average number of visits with that resident was 5.1.

Comparison of the Patient's Perception of the Relationship with the Physician's Perception

Table 1 compares the patient's rating of the relationship with that of the physician. By design, about half (n = 77) of the patients who responded had been identified as being in satisfying relationships by their physician and half (n = 74) as being in problematic relationships. Among the 77 patients who were involved in relationships described as satisfying by the resident, 10% described the relationship as problematic, while 90% described it as satisfying. Among the 74 patients involved in relationships described by the resident as problematic, 23% also viewed the relationship to be problematic (P = .03), and 77% viewed it as satisfying. The majority of patients in both groups described their relationship with their physician as satisfying. Patients in relationships described as problematic by the resident were 3 times as likely to have a similar view of the relationship (odds ratio [OR], 3.0; 95% confidence interval [CI], 2.8 to 3.7).

Table 1.  Comparison of the Residents' View of the Relationship with That of Their Patients
 Physician's Rating of the Relationship
Satisfying (n = 77)Problematic (n = 74)
Patient's rating of the relationship as satisfying, n (%)69 (90)57 (77)
Patient's rating of the relationship as problematic, n (%)8 (10)17 (23)

Patient and Physician Characteristics Associated with the Patient's Perception of the Relationship

Table 2 compares the demographic characteristics, most common diagnosis, perceived health status, and perceived social support of patients who viewed their relationship as problematic with those of patients who viewed it as satisfying. Patients' perception of their social support, their perception of their general health status, and their most frequent clinical diagnosis were characteristics that were significant on bivariate analysis. Of these, patients' perception of their social support was the only factor that retained a significant association with how they perceived their doctor-patient relationship on multivariate analysis. (OR, 4.0; 95% CI, 2.7 to 5.5).

Table 2.  Comparison of Characteristics of Patients Who Viewed Their Doctor-Patient Relationship as Problematic with Those Patients Who Viewed the Relationship as Satisfying
 Patient's Rating of the RelationshipP Value*
Problematic (n = 25)Satisfying (n = 126)
  • *

    P values are based on multivariate analysis.

Mean age, y ±SD56 ± 1559 ± 15.91
Female, n8469.13
Ethnicity, %  .85
 African American2829 
 Latino American3228 
 Asian American02 
Unemployed, %8078.41
Medicaid/Medicare, %8884.90
Completed high school, %9687.35
Average number of visits/y5.55.09
Most common clinical diagnosis, %
 Evaluation of hypertension4847.91
 Back pain2411.14
 Evaluation of diabetes1217.57
 Chest pain81150
Perceived health status, %  .30
 Very good89 
Perceived social support, %  .01

The mean score on the social support component of the Duke Social Support and Stress Scale was 36 (SD = 24). Patients whose scores fell below the 25th percentile (a score of less than 18) were characterized as having low levels of social support. Of the patients who rated the relationship as problematic, 76% also reported low levels of social support, whereas only 16% of patients who reported their relationship as satisfying reported low social support (P = .001).

There was no association between the demographic characteristics of the physician and the patient's perception of the relationship. The physician characteristics that were most closely associated with the patient's perception of the relationship as being problematic were physicians' availability and their ability to handle medical complaints. Seventy-two percent of the physicians who were involved in relationships rated as problematic were rated poor by the patient with regard to their availability as compared to 28% of physicians involved in relationships rated as satisfying (P = .001). Of the physicians who were involved in relationships rated as problematic, 56% were also rated as poor regarding their ability to handle their patients' medical complaints as compared to 44% of physicians involved in relationships rated as satisfying (P = .001).


There have been several studies describing the characteristics of patients who are viewed as troublesome or difficult by their physicians.1–9 This study provides additional insight into the problematic doctor-patient relationship by inviting patients to express their opinions. Furthermore, we looked at a specific type of doctor-patient relationship, in which care is provided predominantly by medical residents. We found that most patients rated their relationship as satisfying. This was not surprising, since it has been found that patients generally report satisfaction with care.10 This finding may reflect a tendency for patients to provide a more “socially desirable” response and report that they are satisfied with the relationship.11 The greatest proportion of patients who rated their relationship as problematic was among patients who were in relationships previously identified as problematic by the resident. Prior studies have also shown that patients who are identified by their physicians as being in difficult relationships are themselves more likely to be dissatisfied with medical care.4,7,8 This finding has implications for conducting research on patient dissatisfaction. Sampling among patients who have been identified by their physicians as being in problematic relationships may be a more sensitive method for identifying patients who are dissatisfied with care.

We know about the demographic characteristics and personalities of patients involved in difficult relationships. However, little is known about interactions between patients' perception of their doctor-patient relationship and their perception of social support. We found 1 study by Carmel that found that social network size was a predictor of satisfaction.18 However, this was done among hospitalized patients, not in an ambulatory care setting. Poor social support emerged as a significant determinant of patients' perception of their relationship. This association remained significant after adjusting for patients' demographic and clinical characteristics as well as their perception of their health status. It is possible that patients with poor social support may have unrealistic expectations of receiving additional support from their physicians. When these expectations are not met, patients may become discontented with the relationship.

Physicians involved in problematic relationships were more likely to be perceived as being unavailable and incapable of handling their patients' medical complaints. Medical residents may have to juggle many tasks while they are seeing their ambulatory care patients, such as responding to the needs of their inpatients, supervising interns and medical students, being on call, and admitting new patients. Patients may be unaware of their physicians' other responsibilities and therefore view them as incapable or unavailable.

Our study has several limitations. First, the results cannot be generalized to all doctor-patient relationships, since we focused on the resident-patient relationship. Second, despite multiple attempts to contact all patients, our response rate was 63%. The demographic characteristics of nonresponders were similar to those of patients who responded. It is possible that there may have been a greater proportion of dissatisfied patients among those who did not respond. If a greater proportion of the dissatisfied patients had responded, we might have been able to identify additional variables that influenced patients' perceptions of the relationship.

The problematic doctor-patient relationship can be especially challenging for physicians in training. Understanding the other side of the story, the patient's perspective, provides additional information about the complexities of the doctor-patient relationship. This study suggests other ways in which the doctor-patient relationship may be enhanced. Physicians should be sensitive to the possibility that patients who perceive inadequate social support may potentially look to their health care providers for support. It might prove beneficial to the relationship if residents were to let their patients know about some of their other responsibilities and that they might not always be available. Patients may also need to be reassured that in the absence of their physicians, there is always a covering doctor. This may potentially assuage potential concerns about physicians' availability. While we did not explore this in our study, patients with poor social support may have some fears of abandonment.

When confronted with problems in the relationship, a patient-centered approach may be helpful. Concato and Feinstein suggest asking 3 questions in assessing satisfaction among primary care patients: “What do you like the most,”“What do you like the least,” and “What one thing would you like to see changed.”25 Periodically asking patients for their opinions of the relationship may be informative and beneficial to both patients and physicians. Asking patients their opinion may stimulate open dialogue about problems in the relationship. Recognizing that patients may be equally frustrated with the care they are receiving may allow physicians to be more introspective and to look at their own role in the relationship. Avoiding labels such as “difficult patient” or “frustrating patient” in this otherwise dyadic relationship removes a negative emphasis on the patient and allows for a more unbiased evaluation of the relationship.


Funding for this research was provided by a Health Services Research Training grant (no. HS00066-06) funded by the Agency for Healthcare Research and Quality and a grant from the Cornell Applied Gerontology Research Institute.