The role of expectations in preferences of patients for a female or male general practitioner

Abstract

Objective

To determine, for five health problems, whether preference for a male or female general practitioner (GP) is related to patient gender, GP gender and/or patient expectations of GP behaviour.

Methods

Cross-sectional study in 14 health centres in Spain, administering a questionnaire to 360 patients. Outcome variables were: preference for male GP, female GP or no preference in consultations for five hypothetical health problems.

Results

Mean age was 47.3 ± 16.5 years, 51% were female. Preference was more frequently expressed by females. Odds ratios (ORs) for a woman preferring a female to male GP ranged from 3 to 508, according to the hypothetical problem, and ORs for a patient with female GP preferring a female GP ranged from 2.8 to 9.1. Patient gender and GP gender had no interactive effect on preferences. Expectations of GP behaviour were related to preferences, except for chest pain. Higher expectations of communication or technical care were associated with greater preference for female or male GP, respectively.

Conclusions

Patient gender and current GP gender are related to preferences in five hypothetical clinical situations and expectations of GP behaviour to preferences in four of them.

Practice implications

Educational strategies are needed to adjust clinical encounters to patients’ preferences.

Introduction

Medicine has always been a profession with a majority of male practitioners, but the presence of women has increased over recent decades, particularly in specialties such as Family Medicine [1]. In-depth studies are therefore warranted into the changes that this growing feminization may produce in doctor–patient relationships and in the organisation of health systems. Previous investigations into clinical practice according to the gender of the general practitioner (GP) have often shown that female GPs conduct longer consultations than their male counterparts and that their care is more focussed on the patient and on communication [2]. These aspects are prioritized by primary care patients, who more often expect their GP to deliver patient-centred care, information, support and longer consultations than to offer technical elements [3], [4], [5].

It has frequently been found that patients, especially women, prefer a GP of the same gender [6], [7], [8], [9]. This preference has been empirically demonstrated in analyses of actual care practice, which showed that female GPs have more female patients in comparison to male GPs [10], [11], [12].

Preferences for a GP are also related to the type of health problem in question, finding a much higher tendency to GP gender preference for genital, social and mental health problems than for other biomedical conditions.

For urogenital problems, the percentage of patients preferring a GP of their own gender reached 32% [9], 63% [7], and even 67% [8], and these proportions were always higher among female than male patients. In obstetrics [13], [14], more women prefer a female than male GP, while in urology [15], there are more women who prefer a female GP than men who prefer a male GP.

For psychological and social problems, a tendency to prefer a female (41%) than a male (9%) GP has been reported [8], and more women more often preferred a female to a male GP, while male patients showed no differences. A wider study of 13 healthcare professions [16] found that female patients had a marked preference for a woman in certain professions (nursing, social work, midwifery and gynaecology) and a somewhat lesser preference for a woman in others (GP or psychiatry). In contrast, male patients only preferred a male surgeon and GP, preferring female nurses and social workers. The above-cited data on actual clinical practice also revealed that female GPs treat more female-specific health problems and psychosocial problems in comparison to male GPs.

Some authors have reported that the patients’ experience of or previous contact with a female GP is related to a greater preference for a female GP [16] or obstetrician [14]. In patients with a female doctor, their preference for a psychosocial orientation in the GP–patient relationship may converge with their experience of their current GP, enhancing their preference for a female GP.

All of the above indicates the complexity of the multifactorial phenomenon of GP gender preference, which varies as a function of the gender of the patient, the gender of the GP, the health problem in question and the gender of previous GPs of whom patients have experience. Previous investigations have reported on some but not all determinants of preferences. We have found no studies that analysed the relationship between the expectations or wishes of the patients for the behaviour of their GP in the consultation and their preference for GP gender.

The present research has the potential to address the set of determinants described above and, using the same sample of patients, to establish the relationships between the preference for the gender of the GP on one side and the type of health problem on the other, while also studying their expectations of GP behaviour. This analysis allows the effect of each factor to be determined, controlling for the remaining factors and endowing the study with high relevance.

We developed three related hypotheses: (a) a female GP would more frequently be preferred by female patients, (b) a higher expectation of GP behaviour oriented towards relational care aspects would increase the preference for a female GP, and (c) experience of receiving care from a female GP would increase the preference for a female GP. The objectives of this study were to determine, for five health problems, whether preference for a male or female GP is related to (1) the gender of the patient, (2) the patient's expectations of GP behaviour, and (3) the gender of the patient's current GP.

A cross-sectional, multicentre study was performed in patients from 14 teaching* healthcare centres in two Andalusian cities (*for Family Medicine post-graduates). Study inclusion criteria were: age ≥ 18 years, visit to GP in previous 12 months and absence of severe psychological or cognitive disability.

Multi-stage random sampling was carried out (alpha = 5%, power = 80%, prevalence = 5%, precision = 5%, sampling effect = 1.20), first in a sample of 30 GPs (15 each gender), and then in a sample of 22 patients per GP (11 each gender), yielding a sample of 660 patients. For budgetary reasons, we had to reduce the sample size by half (n = 360 patients), randomly selected from the GP's patient list. The patients’ records were subsequently reviewed to confirm compliance with inclusion criteria. This reduction in sample size implies a statistical power <60% for some comparisons.

Patient study variables were age, gender, marital status, schooling, employment situation (according to the Spanish National Statistics Institute), social class (by occupation [17]), GP utilization (No. visits in previous 12 months), subjective health, chronic health problems (according to the Spanish National Health Survey); continuity (time with current GP), satisfaction with GP (five possible responses), and expectations of GP behaviour in the consultation. Following the taxonomy of Kravitz, expectations were defined as “wishing that something will happen” in the visit, which is different from “expecting that something is likely to occur” or “asking the GP to do something” [18], [19].

The 18 items in the scale of expectations on GP behaviour were prepared by our group or adapted from Little et al. [5], Grol et al. [3] and Williams et al. [4], and they were repeated for the five problems under study. A pilot study with 21 patients led to the elimination of questions that did not appear to discriminate, producing a definitive 12-item version of the questionnaire. A 5-point Likert scale was used for responses, ranging from “Very Important” to “Not at all important”.

According to published reports, the expectations of patients vary, with relationship aspects being more important on some occasions [20] and technical skills on others [21]. Given the complexity and instability of expectations, Staniszewska and Ahmed [22] recommended their study in homogeneous groups of patients with a given disease. For this reason, as previously reported [23], we selected five health problems and analysed the factorial structure of the expectations for each of them in order to identify the empirical dimensions of the expectations of the patients according to the type of health problem in question.

Expectations were evaluated for five hypothetical health problems (clinical scenarios) that are directly or indirectly known by a large majority of the general population, using clinical vignettes. We selected one urogenital problem (abnormal discharge from penis or vagina), two psychosocial problems (one psychological: depression/sadness, one social: severe family problem), for which previous studies have found a strong relationship with GP gender. We also selected two biomedical problems (one severe: strong chest pain, one mild: cold with fever), for which a GP gender preference is reportedly much less frequent. This portfolio of hypothetical conditions was selected in order to analyse whether the effect of the type of problem on preference is as marked as has been claimed.

The factorial and homogeneity analyses of the expectation scales showed five differentiated empirical structures [23]: three factors for chest pain (communication, experience of disease and technical interventions) (Cronbach's alpha = 0.843), three for discharge (experience of disease-specialist technical interventions, communication-GP technical interventions, and communication) (Cronbach's alpha = 0.871), four for depression (experience of disease-communication, GP technical interventions, listening, and referral to specialist) (Cronbach's alpha = 0.761), three for family problem (communication-experience of disease, technical interventions, and affect on life) (Cronbach's alpha = 0.881), and five factors for cold (specialist technical interventions, GP technical interventions, experience of disease, communication, and listening) (Cronbach's alpha = 0.921).

The dependent variable was “patient preference for a male or female GP”, measured by asking the following question (repeated for all five problems): “If you had a strong chest pain/cold with fever/abnormal genital discharge/depression-sadness/severe family problem, who would you prefer to be seen by?” There were always three response options: female doctor, male doctor or no preference. In our study, preference is understood as an evaluation made by the patient of an aspect of care that might influence their choices for their healthcare, in this case selection of the gender of their GP [24].

Patients were interviewed at home (February–June 2003) with no previous notification. They were considered lost to the study if not found at home on three visits at different times, and these patients and those refusing to respond were replaced with the next patient on the same GP's list. Interviewers underwent specific training and received an interview manual in order to minimise information bias.

Descriptive, bivariate and polytomous logistic regression analyses were performed. The comparison of interest was female versus male GP, with the latter as reference category, forcing the second comparison between no preference and preference for male GP. We obtained two regression models for preferences in each situation, controlling for the above-mentioned confounding patient and GP variables. p < 0.01 was considered significant. STATA 9.0 and SPSS/PC 11.5 programmes were used for data analyses.