Sexual orientation disclosure in health care: a systematic review

Abstract

Background

Significant health disparities between sexual minority individuals (that is, lesbian, gay, bisexual, or transgender [LGBT]) and heterosexual individuals have been demonstrated.

Aim

To understand the barriers and facilitators to sexual orientation (SO) disclosure experienced by LGBT adults in healthcare settings.

Design and setting

Mixed methods systematic review, including qualitative, quantitative, and mixed methods papers following PRISMA guidelines.

Method

Study quality was assessed using the Mixed Methods Appraisal Tool (MMAT) and a qualitative synthesis was performed. Studies were included if their participants were aged ≥18 years who either identified as LGBT, had a same-sex sexual relationship, or were attracted to a member of the same sex.

Results

The review included 31 studies representing 2442 participants. Four overarching themes were identified as barriers or facilitators to SO disclosure: the moment of disclosure, the expected outcome of disclosure, the healthcare professional, and the environment or setting of disclosure. The most prominent themes were the perceived relevance of SO to care, the communication skills and language used by healthcare professionals, and the fear of poor treatment or reaction to disclosure.

Conclusion

The facilitators and barriers to SO disclosure by LGBT individuals are widespread but most were modifiable and could therefore be targeted to improve healthcare professionals’ awareness of their patients’ SO. Healthcare professionals should be aware of the broad range of factors that influence SO disclosure and the potential disadvantageous effects of non-disclosure on care. The environment in which patients are seen should be welcoming of different SOs as well as ensuring that healthcare professionals’ communication skills, both verbal and non-verbal, are accepting and inclusive.

INTRODUCTION

Significant health disparities between individuals identifying as part of a sexual minority (that is, lesbian, gay, bisexual, or transgender [LGBT]) and heterosexual individuals have been demonstrated internationally.^1,2 In the UK, sexual orientation (SO) is a protected characteristic under the Equality Act (2010), which requires public services to promote and demonstrate equality for LGBT people. A large component of proving compliance with this mandate is monitoring SO, which is currently poorly done in the UK. National estimates of the adult LGBT population range from 1.7%³ to 9.9%,⁴ although the validity has been questioned.⁵ This has been recognised as a significant issue, and NHS England has worked with the LGBT Foundation and National LGB&T Partnership to implement an SO monitoring information standard from April 2017.⁶

Health disparities between heterosexual and LGBT people are still seen in mental health, with higher rates of anxiety and depression, self-harm, and suicide^1,7–11 among the LGBT community, as well as in physical health. A recent UK-based review reported increased rates of some malignancies in the LGBT community, mixed diabetes rates, and higher rates of substance abuse, including binge drinking and smoking.¹ Differences between sexual minority groups have also been reported, showing poorer mental and physical health in bisexual people of both sexes,^8,9,11 as well as higher rates of high-risk health behaviours, such as smoking and excess alcohol intake.^1,7,8,10 It has been noted that robust evidence comparing the different groups that make up the LGBT community is lacking,¹ particularly in reference to transgender, queer, and intersex persons.

The most prominent theory for differences in health by SO is minority stress.^12,13 This hypothesises that a combination and accumulation of internal and external stressors (such as stigma and victimisation, and the distress felt in response to stigma and concealment of one’s SO) interact to overcome an individual’s ability to cope, resulting in psychological and physical disease.¹³ A further theory is fundamental causes, which posits that advantaged groups in society have the skills and resources necessary to minimise risk of disease, as well as to harness the appropriate health resources to lessen the consequences of disease, should it occur.¹⁴ A Swedish study has presented support for the fundamental cause theory applicable to the LGBT community, describing increased rates of high-preventable diseases — such as ischaemic heart disease, chronic obstructive pulmonary disease (COPD), and lung cancer — in LGBT people, compared with heterosexual people.¹⁵

How this fits in

Significant health disparities exist between sexual minority (that is, lesbian, gay, bisexual, or transgender [LGBT]) and heterosexual individuals. Disclosure of sexual orientation (SO) in health care links to both the minority stress and fundamental cause theories in the context of accessing appropriate services, and is therefore likely to be a contributing factor in these health differences. Incorporating more LGBT-specific knowledge and communication skills into undergraduate medical education is essential in aiding SO disclosure. Altering the healthcare environment, such as displaying signs or symbols that convey an accepting atmosphere — for example, a rainbow symbol or the Human Rights Campaign logo — may also help.

Disclosure of SO in health care links to both the minority stress and fundamental cause theories in the context of accessing appropriate services, and is therefore likely to be a contributing factor in the health differences. In line with this, a recent British review found that many LGBT people are reluctant to disclose their SO, and will sometimes delay care due to fear of disclosure, even in the face of inappropriate or less appropriate care.¹⁶ The purpose of this review was to investigate the barriers and facilitators to SO disclosure in health care by LGBT adults, with the aim of identifying factors that can be easily modified in healthcare education and practice to improve disclosure, and therefore ensure provision of appropriate care.

METHOD

Search strategy

A search of eight databases (AMED, CINAHL, Embase, MEDLINE, PsycINFO, RCNi, ScienceDirect, and Web of Science) was conducted in March 2017. Terms were chosen to include all standard gender categories and minority SOs, focusing on SO disclosure in healthcare settings or to a healthcare professional. The final search conducted was: ((disclos* OR reveal* OR openness) AND (lgb* OR gay OR bisexual OR lesbian OR msm OR wsw OR homosex*) AND (health* OR care OR consult*)). The authors excluded all editorials, commentaries, reviews, and conference abstracts. Only articles published after 2000 were considered to ensure recent barriers and facilitators were captured, and only those in English were included.

Inclusion and exclusion criteria

Participants were aged ≥18 years and samples contained at least some self-identified as LGBT. Furthermore, only studies that displayed data provided by the participants on the barriers and/or facilitators to disclosure (or non-disclosure) of SO to a healthcare professional were included. Studies that did not specify disclosure to a healthcare professional, or those outside a healthcare setting, were excluded. Although the authors recognise that transgender is a gender identity rather than an SO, they have included transgender as they were unable to disaggregate transgender from LGB data.

Study selection and data extraction

The process of systematic review is summarised using the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA)¹⁷ (Figure 1). Data were extracted using a proforma, followed by qualitative analysis.

Figure 1.

Preferred reporting items for systematic review and meta-analysis flow diagram for the inclusion of studies reporting barriers and facilitators to sexual orientation (SO) disclosure in healthcare.

Assessment of study quality

The Mixed Methods Appraisal Tool (MMAT) ¹⁸ was used to assess methodological quality. Two screening questions and four criteria for assessment were applied to each study, scoring sampling, measurement, analysis, and limitation consideration. This gave a score ranging from 0% (no criteria met) to 100% (all four criteria met) for each study, allowing one robust score to be used for multiple study types. Quality assessment was carried out by three assessors. Kappa scores were calculated to assess inter-rater variability.¹⁹

RESULTS

Studies identified

From 2603 records, 31 studies met the inclusion criteria (Figure 1; Table 1). Six studies presented data relevant to disclosure solely in a primary care setting, three in oncology, three in military medical settings, and one each in mental health and a home care setting. Eleven studies did not state or did not specify a precise healthcare setting but instead presented data from generic health settings, and six presented data from a variety of settings. In total, 2442 participants were included across the 31 studies identified for review.

Table 1.

Characteristics of studies included in the review (presented in chronological order)

Author (publication year) country	Healthcare specialty	Study population	Selected sample characteristics (age, ethnicity, education level)	Sample size	Study design	Recruitment method	MMAT score, %
Barbara46 (2001) US	Not stated	Lesbian women	Age range 24–65 years; 69% white; 90% some college education	32	Qualitative, focus groups	Purposive, community-based	75
Beehler21 (2001) US	Primary care	Gay men	Median age 38 years (range 25–52); 82% white; 91% some college education	11	Qualitative, interview	Snowball	50
Stein29 (2001) US	Range — 77% primary care	Men and women of any sexual orientation	Mean age 45 years (range 19–83); 76% white; 80% at least college education	575	Quantitative descriptive, survey	Convenience, snowball; community-based	25
Boehmer32 (2004) US	Oncology	Lesbian and bisexual women	Mean age 49 years (range 26–67); 94% white; 97% at least college	39	Qualitative, interview	Purposive, community-based snowball	75
Clover48 (2006) UK	Not specified	Gay men	Age range 60–75 years; 100% white	10	Qualitative, interview	Purposive, community-based	50
McDonald50 (2006) Canada	Not specified	Lesbian women	Age range 26–56 years; 100% white; 73% at least some college education	15	Qualitative, interview	Purposive, community-based	25
Bjorkman40 (2007) Norway	Primary care	Lesbian women	Mean age 41 years (range 28–59); all white; all well educated	6	Qualitative, focus groups	Convenience, online	75
Mulligan35 (2007) Australia	Not specified	Lesbian and bisexual women	Age range 20–71 years	47	Qualitative, interview	Purposive, online, and community-based	50
Adams20 (2008) New Zealand	Primary care	Gay men	Not specified	50	Qualitative, focus groups	Not specified	50
Bjorkman31 (2009) Norway	Any, majority primary care	Lesbian women	Age range 18–≥60 years (68% aged 20–39); 87% Norwegian native; 67% some college education	121	Qualitative, online	Convenience, online questionnaire	75
Politi36 (2009) US	Not specified	Women of any sexual orientation	Mean age 55 years; 98% white; 73% at least college education	40	Qualitative, interview	Convenience, community-based	50
Daley41 (2012) Canada	Mental health	Lesbian women	Age range 20–58 years; 83% white	18	Qualitative, interview	Purposive, healthcare, and community	75
Biddix44 (2013) US	Military	Gay and bisexual men	Age range 18–47 years (56% 18–27); 86% white; 91% some college education	30	Quantitative, descriptive survey	Convenience, online	0
Johnson22 (2014) US	Not specified	Non-heterosexual women	Mean age 20 years (range 18–23); 77% white; all university students	9	Qualitative, interview	Purposive, online	75
Koh23 (2014) Australia	Primary care	Lesbian, gay, bisexual, or transgender men and women	Modal age 20–29 years (range 18–≥60)	99	Qualitative, online questionnaire	Purposive, online, and print	75
Sharek38 (2014) Ireland	Range	Lesbian, gay, bisexual, or transgender men and women	59% aged 55–59 years	144 survey + 36 interview	Mixed methods	Convenience, print/events/online	50
Sherman28 (2014) US	Military	Lesbian, gay, bisexual, or transgender men and women	>50% aged 40–59 years; 84% white	58	Mixed methods	Convenience, healthcare, online, and community	0
Wirtz49 (2014) Malawi	Range	Men who have sex with men	Not specified	8	Qualitative, interview	Responder-driven, purposive	75
Law34 (2015) Canada	Primary care	Lesbian, gay, bisexual, or transgender men and women	Mean age 32 years; 91% university education	12	Qualitative, interview	Purposive, snowball	75
Marques24 (2015) Portugal	Not specified	Lesbian women	Mean age 37 years (range 21–63)	30	Qualitative, interview	Snowball	75
Mattocks25 (2015) US	Military	Lesbian women	Age range 41–50 years; 35% white, 30% Hispanic, 15% African American	20	Mixed methods	Purposive, print, and healthcare	0
Quinn42 (2015) US	Not specified	Men and women of any sexual orientation	Not specified	632	Mixed methods	Purposive, online	50
Underhill39 (2015) US	Range	Men who have sex with men	Median age 27 years (male sex workers); median age 39 years (MSM); 76% white; 40–50% college education	56	Qualitative, interview	Convenience, snowball	75
Fish33 (2016) UK	Oncology	Lesbian, gay, or bisexual men and women	Mean age 54 years (range 41–71)	15	Qualitative, interview	Purposive, community, and web-based	75
Furlotte47 (2016) Canada	‘Home care’	Lesbian, gay, bisexual, or transgender men and women and their partners	Mean age 64 years (range 39–75); 96% white	24 (12 couples)	Qualitative, individual and paired interviews	Web-based, purposive, snowball	75
Legere45 (2016) Canada	Oncology	Lesbian and bisexual women	2 in 20s, 4>40 years; 28% black	7 (6 patients, 1 HCP)	Qualitative, interview	Purposive, print, and online	100
Munson26 (2016) New Zealand	Primary care	Lesbian and bisexual women	Age range 23–47 years; 83% higher education	6	Qualitative, interview	Purposive, snowball	100
Roller27 (2016) US	Not specified	Lesbian and bisexual women	Mean age 41 years (range 21–59); all white; 67% college degrees	13	Qualitative, interview	Purposive, online	75
Venetis30 (2017) US	Not specified	Lesbian, gay, bisexual, or transgender men and women	Mean age 28 years (range 21–44); 66% white	24	Qualitative, interview	Purposive, community and online, snowball	100
Wanyenze43 (2016) Uganda	Not specified	Men who have sex with men	50% aged 21–25 years	85 + 61 key informants	Mixed methods	Purposive, snowball, responder-driven	0
Rose37 (2017) International	Oncology or primary care	Gay and bisexual men	Mean 64 years; 67% white	124 + 21 partners	Qualitative, interview	Purposive, online, and postal	50

HCP = healthcare professional. MMAT = Mixed Methods Appraisal Tool. MSM = men who have sex with men.

Data synthesis

The barriers and facilitators identified are presented in four overarching themes (Box 1).

Box 1. Facilitators and barriers to sexual orientation disclosure in health care.

Facilitators	References	Barriers	References
Moment of disclosure		Moment of disclosure
Communication skills of healthcare professional		Communication skills of healthcare professional
Response to a direct question	21,23,26–28,31,33,34,36,37,39,41,44,45	Response to a direct question	25,27,30,41
Inclusive language	30,34,36,40,41	Heteronormative language	26,34
Open body language	23,34,36,41–43	Closed body language	41
		No opportunity in conversation	33,37
Relevant to care	20–30	Irrelevant to care	21,23–25,27–29,31–39
Written disclosure	22,35,39,41,42,46	Written disclosure	26,36
Confronting heteronormative assumptions	21,27,29–31,33,40,47	Conforming to heteronormative assumptions	21,26,45,46
Perceived outcome of disclosure		Perceived outcome of disclosure
Patient–provider confidentiality	22	Breach of confidentiality	20,24,29,34,37,39,43,46,47,49
Documented on medical record	24	Documented on medical record	24,25,28,29,46
Good/open healthcare professional response	32	Poor healthcare professional response	23,24,29,30,37–41,43,46,48,49
		Embarrassment	31,33,37,39,46
		Discrimination
		Poorer care	23,26–29,32,40,43,45,47,48
		Loss/impact on job	25,43
		Loss of benefits	25,28
		Criminalisation	43,49
Healthcare professional factors		Healthcare professional factors
Perceived accepting of LGBT	32,34,39,40,43,46	Perceived non-accepting of LGBT	32,37,38,48
Long relationship with patient	23,39,47	Long relationship with patient	34
Short relationship with patient	39	Short relationship with patient	46
Gender	36,38,39	Ill-informed of LGBT issues	20,31,36,46,48
LGBT	20,25,32,38,41,46,48
Environmental factors		Environmental factors
Location/setting	35	Location/setting	32,35,38,39,44
Accepting visual cues	23,26,39,41,42,48,50	Religious icons	23
Supportive community	32,41	Unsupportive community	43,49

LGBT = lesbian, gay, bisexual, or transgender.

The moment of disclosure

Twenty studies commented on patients’ beliefs of the relevance of SO to health care as both a barrier and facilitator to disclosure; people who thought it was relevant were more likely to disclose,^20–30 whereas those who thought it was irrelevant were less likely to reveal their identity.^{21,23–25,27–29,31–39} One participant felt the need to disclose to enable their healthcare professional to provide ‘more focused advice’ ²⁴ and another thought their ‘gayness to be highly relevant to [their] health needs’.²¹ Others asked ‘what’s [my SO] got to do with, you know, my toe hurting?’ ²⁸ and felt ‘ [SO] would only be important if a problem was discovered’.³⁶

Communication factors, such as using inclusive language^{30,34,36,39–41} and open, welcoming body language,^{23,34,36,41–43} were seen as facilitators to disclosure whereas the opposites — closed-off or unfriendly body language⁴¹ and heteronormative language,^26,34 such as using a male pronoun to identify a female patient’s partner, and vice versa — were viewed as barriers. There were mixed opinions on the merits of using direct questions to explore a patient’s SO. The majority of participants appreciated being asked and felt this was a good way to facilitate open communication between patient and provider,^{21,23,26–28,31,33,34,36,37,39,41,44,45} but a small number did not agree.^25,27,30,41 There were similarly mixed views of the benefits of patient registration forms to document SO. Some described their delight at finding a registration form that included their SO as an option,³⁵ whereas many felt their SO was not accommodated by the options presented.^26,36 Most described these types of written disclosure as a facilitator to disclosure,^{22,35,39,41,42,46} but only if they were adapted to be more inclusive and depict a broad spectrum of SOs.²²

The final barrier at the moment of disclosure was the patient’s response to heteronormative assumptions. This was most commonly identified in the context of contraception and sexual health, with the giving of only heterosexually appropriate advice.^27,30,31,45

Perceived outcome of disclosure

Fear of discrimination, including receiving poor or unequal care,^{23,26–29,32,40,43,45,47,48} having a negative impact on their career^25,43 or benefits,^25,28 as well as criminalisation,^43,49 were all cited as reasons not to disclose. In addition, many participants were hesitant to disclose for fear of a negative personal reaction from their healthcare professional,^{23,24,29,30,37–41,43,46,48,49} or feeling embarrassment or humiliation after disclosure.^{31,33,37,39,46} Many participants cited concerns of breaches in patient–provider confidentiality^{20,24,29,34,37,39,43,46,47,49} that would lead to non-clinical staff,⁴⁷ their family and friends,³⁴ or the wider community^43,49 discovering their SO as reasons not to disclose. Similarly, documentation of SO in medical records was seen as a barrier to disclosure.^{24,25,28,29,46}

Healthcare professional factors

The majority of patients were more likely to disclose to a healthcare professional with whom they had a long relationship.^23,39,47 Seven studies reported an increased likelihood of disclosure if the healthcare professional was themselves a member of the LGBT community.^{20,25,32,38,41,46,48} Although having a heterosexual healthcare professional was not seen as a particular barrier to disclosure, a healthcare professional being perceived as accepting of the LGBT community, or of their patient being LGBT, was a significant facilitator.^{32,34,39,40,43,46}

Environmental factors

Some participants preferred to disclose their SO in sexual health clinics rather than to their primary care provider.³³ On the other hand, military⁴⁴ and religious-affiliated³² settings were seen as impeding disclosure, as was care conducted in a group treatment setting.³⁹ Most notably, seven studies commented on visual clues in the healthcare setting that facilitated disclosure.^{23,26,39,41,42,48,50} These included seeing leaflets, stickers, and posters that were deemed LGBT friendly, such as the Human Rights Campaign logo or a rainbow sign.^41,42,48 Religious symbols or icons displayed in the healthcare professional setting were barriers.²³

DISCUSSION

Summary

In the UK, it is estimated that only half of lesbian and gay people are out to their GP, with disclosure rates lower in bisexual people.¹⁶ The authors have found that the factors promoting or discouraging patient SO disclosure in health care are widespread and varied. The most commonly cited factors were associated with the patient–provider interaction, which may provide useful targets to improve disclosure rates. Factors that were deemed to either enhance or reduce SO disclosure among females were having SO documented in their medical record and using written forms as a means of disclosing SO, as well as the type of language used during a consultation. Perhaps the use of prompts to aid disclosure, such as having a partner, a written form, or picking up on clues from the healthcare professional’s speech, are more important to LGBT women than men as they may be more commonly assumed heterosexual, particularly in discussing their reproductive health,^30,51 and are less frequently asked directly about their SO.²⁸

Although almost all studies were conducted in countries where homosexuality is legal, two were not. In both of these, barriers to disclosure were almost exclusively explored: commonly, the effect of an unsupportive community, fears of discrimination, and breaches in confidentiality were described by participants. They were, unsurprisingly, the only two studies to mention criminalisation as a barrier. Although the factors explored were often extreme, ranging from not being treated by their healthcare professional at all, to the police being informed of the participant’s SO, and fears of being ostracised from their community, they were echoed to a lesser extent in studies based in other countries.

Strengths and limitations

Although this is the first review to include participants that are both men and women, as well as participants from any sexual LGBT subgroup, there are some limitations. The MMAT has shortcomings. Although it allows the authors to assess different study types with one tool, they often found it difficult to assess the methodological qualities of each study without assessing the quality of reporting. Further, the authors found the MMAT criteria to be fairly crude measures of quality, particularly for qualitative studies. The quality assessment was not taken into account when extracting data from each study, with all the evidence being treated equally. Additionally, most of the mixed methods studies had particularly weak evidence from the quantitative branch of the study. The richest and most appropriate data were extracted from the qualitative arms.

The studies included for review also have limitations. Sampling the LGBT community is recognised as difficult due to the hidden nature of the population. The authors recognise participants need to have disclosed their SO before being recruited to studies, so may not have the same barriers and facilitators to disclosure as those who had not disclosed at all. Furthermore, the participants from each study were largely homogenous, comprising mostly well-educated, white, middle-aged people, who are the groups most likely to disclose their SO.^52–55

Comparison with existing literature

Studies with only correlates of SO disclosure were excluded as they were outside the remit of this review. They do include, however, important information on the effects of patient sex, age, ethnicity, and SO on disclosure. For example, LGBT people who are from ethnic minorities,^52–55 or identify as bisexual,^{52,53,55–57} or do not have a college education,^53,54,58 or have a low income^53,54 are less likely to disclose their SO to a healthcare professional. There is mixed evidence for the effect of patient age^53,54,58 and sex^16,52,59 on disclosure. These are important factors to consider when implementing interventions in terms of targeting population groups.

Although useful to enhance the authors’ understanding of demographics and disclosure, the quantitative data also support the predominantly qualitative findings. For example, a recent study from Canada found that higher levels of self-esteem, having a partner, and higher levels of social support from friends were significantly associated with healthcare professionals knowing a patient’s SO, whereas participants with previous experiences of discrimination and higher levels of internalised homonegativity were less likely to discuss LGBT-related health issues with their healthcare professionals.⁶⁰

Implications for research and practice

Although some of the factors identified in this study are fixed, some could be targeted to minimise the barriers to disclosure. Five of the studies in this review commented on healthcare professionals’ lack of LGBT-specific knowledge as a barrier to disclosure. This problem stems from the beginning of medical education, with one study noting a median of five LGBT-dedicated curriculum hours in US medical schools,⁶¹ and another study showing medical students in the UK lack confidence in the use of LGBT-specific health terms and their ability to locate LGBT-specific health information.⁶² Incorporating more LGBT-specific knowledge and communication skills into undergraduate medical education is essential to ensure that future healthcare professionals are armed with the tools they need to help their future patients disclose their SO, and then provide them with appropriate care and advice. The responsibility for medical education does not just sit within the undergraduate realm: there should be increased presence of LGBT-specific issues and appropriate communication tools in postgraduate curricula also.

At an institutional level, the design of healthcare settings should take into account the needs of LGBT patients. There are some changes that are easily implemented and inexpensive, including displaying signs or symbols that convey an accepting atmosphere, such as a rainbow symbol or the Human Rights Campaign logo, while others may take more time. It is important to ensure, however, that any healthcare setting changes are congruent with the beliefs of the healthcare professional working within them. A key intervention is the production of patient information leaflets that are accepting of the LGBT community and that consider the differing needs of LGBT individuals compared with heterosexual individuals, providing LGBT-specific information when necessary.

Individual healthcare professionals should be aware of the differing physical and psychological needs of the LGBT community and remain open minded regarding their patients’ SO. The authors encourage all healthcare professionals to reflect on their use of language, keeping an eye out for heteronormative phrases and assumptions, as well as those that may be inhibiting their patients’ ability to disclose, and consider using alternative terms. The most common example of this is referring to a patient’s partner as ‘he’ or ‘she’, rather than asking whether they are male or female or going further to ask whether the patient is, or ever has been, involved in a relationship with men, women, or both. Healthcare professionals should also consider asking questions about each patient’s SO in their daily practice, using open and accepting language. Further investigation into issues surrounding disclosure from a healthcare professional perspective would also provide a fuller understanding of the complexities surrounding SO disclosure in health care.

Ideally, robust population-level studies that include an accurate portrayal of the breadth encompassed within LGBT are needed. The current SO monitoring question in the UK has only five possible answers (heterosexual, gay/lesbian, bisexual, other, prefer not to say), which does not display the full spectrum of orientations and focuses only on sexual identity rather than attraction or behaviours. For example, an alternative means of monitoring those who describe themselves as ‘other’ would allow a much richer, and much needed, analysis of the population.

Funding

None given.

Ethical approval

Not applicable.

Provenance

Freely submitted; externally peer reviewed.

Competing interests

The authors have declared no competing interests.

Discuss this article

Contribute and read comments about this article: bjgp.org/letters