Measuring the psychosocial impact of population-based prostate-specific antigen testing for prostate cancer in the UK

Authors


Lucy A. Brindle, School of Nursing and Midwifery, University of Southampton, Nightingale Building, University Road, Southampton, SO17 1BJ, UK. e-mail: L.A.Brindle@soton.ac.uk

Abstract

OBJECTIVE

To evaluate the psychosocial impact of participation in a population-based prostate-specific antigen (PSA) testing programme, akin to screening, and to explore the relationship between urinary symptoms reported before PSA testing and the response to the subsequent PSA result.

PATIENTS AND METHODS

This prospective questionnaire study was nested within the case-finding component of the ProtecT (prostate testing for cancer and treatment) feasibility study (ISRCTN20141297). Men aged 50–69 years from 18 general practices in three cities in the UK completed the Hospital Anxiety and Depression Scale (HADS), the Short Form-12 (SF-12) Health Survey, and the International Continence Society ‘male’ (ICSmale) questionnaires before giving consent for a PSA test in a community clinic (baseline). Men with an ‘abnormal’ PSA result returned for further investigation (including biopsy) and repeated these questionnaires before biopsy.

RESULTS

At baseline, study participants had similar levels of anxiety and depression to the general male population. There was no increase in the HADS scores, or reduction in the SF-12 mental health component summary score, on attendance at the biopsy clinic after receiving an ‘abnormal’ PSA result. Urinary symptoms were associated with levels of anxiety and depression before receiving a PSA result (baseline), but were not associated with anxiety and depression at biopsy independently of baseline scores. Therefore changes in anxiety or depression at biopsy did not appear to differ between those with and without urinary symptoms.

CONCLUSIONS

This study confirms the findings of other studies that the deleterious effects of receiving an abnormal PSA result during population screening are not identified by generic health-status questionnaires. Comparisons with outcomes of studies measuring cancer-specific distress and using qualitative research methods raise the question of whether a prostate cancer screening-specific instrument is required. However, a standardized measure of anxiety identified differences at baseline between those who did and did not report urinary symptoms. These findings suggest that it might be advisable to better inform men undergoing PSA testing about the uncertain relationship between urinary symptoms and prostate cancer, to minimize baseline levels of psychological distress.

Abbreviations
ProtecT

prostate testing for cancer and treatment feasibility study

HADS

Hospital Anxiety and Depression Scale

SF-12(-MCS)(PCS)

12-Item Short-Form Health Survey (mental-health component summary) (physical-health component summary)

PTC

PSA test clinic

ICSmale

ICS ‘male’ questionnaire

(HR)QoL

(health-related) quality of life

PLCO

Prostate, Lung, Colorectal and Ovarian Cancer screening trial.

INTRODUCTION

Screening for prostate cancer using the PSA test is not advocated at a population level in the UK, but the use of the test amongst asymptomatic men in primary care increased significantly between 1999 and 2002 [1]. In a change of emphasis in health policy, The National Health Service Prostate Cancer Risk Management Programme advocated ‘informed choice’ for men requesting PSA tests [2,3], which is likely to have further increased the use of the test since 2002. Using age-based thresholds of PSA levels to indicate abnormality, ≈ 10% of men aged 50–70 years will have a raised PSA level, of whom ≈ 70% will be false-positives [4].

Future decisions about the introduction of population screening are likely to involve an assessment of the balance of the possible benefits and harms involved. The anticipated benefits of prostate cancer screening would result from a reduction in further mortality and possibly the reassurance provided by a ‘normal’ test. Harm could result from the diagnostic testing itself (the biopsy), over-detection and treatment, and any reduction in mental well-being caused by involvement in screening. Population-based screening for prostate cancer is currently being evaluated in randomized clinical trials in Europe. Data from the Rotterdam trial (a participant in the European Randomised Study of Screening for Prostate Cancer) failed to detect any relevant changes in health status or anxiety during the screening procedure [5]. Evidence relating to the psychological impact of PSA testing in a UK context is lacking, and it is important that men’s responses to prostate cancer screening are better documented and understood if individual and policy decisions are to be effectively supported and informed.

Despite a lack of evidence that men with LUTS are at greater risk of prostate cancer than asymptomatic men of the same age [6], men attending for investigation of LUTS in the UK were found to express fears about prostate cancer in relation to their symptoms [7]. LUTS are common; 20% of men aged >55 years in the UK report moderate or severe symptoms [8] and it is possible that men with urinary symptoms are more likely to respond to the offer of a PSA test. Men’s level of psychological distress when attending a PSA test clinic (PTC) and their response to attending for prostatic biopsy after an ‘abnormal’ PSA result might be mediated by the presence of such symptoms.

To evaluate the psychological impact of participation in PSA testing, we conducted a prospective study nested within the case-finding component of the ProtecT (prostate testing for cancer and treatment) Study [4]. This prospective study also provided the opportunity to explore the relationship between levels of urinary symptoms reported at baseline and levels of psychological distress before and after an abnormal PSA test result.

SUBJECTS AND METHODS

In all, 15 151 men (aged 50–69 years) from 18 general practices in three UK centres were invited to attend a nurse-led community clinic (a PTC) to consider having a PSA test as part of case-finding for a randomized controlled trial of treatment for localized prostate cancer (The ProtecT Study, [4]). On attendance at the clinic, men were asked to complete a measure of mental health (the Hospital Anxiety and Depression Scale, HADS), [9] a generic measure of health status (the 12-Item Short-Form Health Survey, SF-12 [10]), and a measure of five LUTS: frequency, nocturia, urgency, urge incontinence and hesitancy (the ICS ‘male’ questionnaire, ICSmale[11]). Men eligible for study participation (no clinical history that would make them unsuitable for any one of the three main treatments: radical surgery, active monitoring or radical radiotherapy), were offered a PSA test (Fig. 1). Initially age-specific PSA level thresholds were used (50–59 years, ≥ 3 ng/mL; 60–69 years, ≥ 4 ng/mL), but on the basis of research evidence this was changed after 12 months to a uniform threshold (≥ 3 ng/mL). Men whose PSA level was above the relevant threshold were told that they had a ‘higher than normal’ PSA level and were invited for biopsy. Men attending for a biopsy completed repeat SF-12, HADS and ICSmale questionnaires, and a sexual function measure, at the clinic before the appointment [12]. Men found to have localized prostate cancer after biopsy were offered randomization to one of three treatment arms [4].

Figure 1.

Flow chart detailing events (squares) and timing of data collection (horizontal arrows) for those invited to have a PSA test. The baseline health status assessment occurred in the waiting room of the PTC before counselling about the implications of having a PSA test. The biopsy health status assessment occurred in the waiting room before biopsy. Biopsy was 2–6 weeks after an ‘abnormal’ PSA result.

The HADS is a 14-item scale with seven items relating to depression and seven relating to anxiety during the previous week. There is a maximum score of 21 on each of the depression and anxiety subscales [9]. Missing HADS values were replaced with the individual’s mean for that scale if there were two or fewer missing values. If more than two items were missing, the scale score was not calculated. The SF-12 consists of 12 items relating to physical functioning, bodily pain, general health, role limitations due to physical problems, vitality, social functioning, mental health and role limitations due to emotional problems. The SF-12 mental-health component summary (SF-12-MCS) and physical-health component summary (SF-12-PCS) scores were calculated using a standard scoring algorithm [13]. Scores potentially range from 0 to 100, with a higher score indicating better health. Missing values were imputed by the SF-12 Health Survey scoring software, which uses missing data estimation algorithms to yield unbiased estimates of the PCS and MCS scores [14].

Each of five urinary symptoms: frequency of voiding, nocturia (getting up in the night to urinate), urgency (having to rush to urinate), hesitancy (delay before starting to urinate) and urge incontinence (leakage of urine before getting to the toilet), was assessed using items from the ICSmale questionnaire. Responses were re-categorized to generate a binary classification of either ‘having’ or ‘not having’ the urinary symptom: a frequency of <3 h between voids, nocturia as getting up more than once per night to urinate; urgency, hesitancy and urge incontinence as any occurrence of the symptom.

The characteristics of the study group (those who returned for biopsy and had complete HADS, SF-12 and ICSmale data) were compared descriptively with those of eligible PTC attenders who consented to a PSA test and PTC attenders with a higher than normal PSA level. Logistic regression models were used to investigate differences in the likelihood of having a raised PSA level between those with and without urinary symptoms, adjusting for age.

Measures of central tendency and the 10th−90th percentile of anxiety, depression and SF-12-MCS scores were reported at baseline and at biopsy for the study group. Differences in scores between the baseline and at biopsy were evaluated using paired t-tests. The relationship between the change in HADS from the baseline to biopsy and the mean level during the study was investigated using graphical and regression methods [14,15]. The relationship between age and HADS scores was explored by investigating the trends across four (5-year) age bands for the mean anxiety and depression scores at baseline, and for the mean change in scores at biopsy.

Linear regression models were used to compare depression and anxiety scores at baseline, scores at biopsy and scores at biopsy adjusted for baseline scores, for those with and without individual urinary symptoms, adjusting for age. Multivariate linear regression models, including all urinary symptoms, were also adjusted for the baseline SF-12-PCS score to explore the possibility that general health status was a confounding variable in the relationship between urinary symptoms and HADS scores.

RESULTS

Of men with an ‘abnormal’ PSA level who were eligible for inclusion in the study, 770 (90%) were known to attend for biopsy. The mean SF-12-MCS (53.3) and HADS scores (anxiety 4.8; depression 2.9) at baseline for these men did not differ from those not attending for biopsy, at 52.5, 95% CI (50.3, 54.8), 4.8 (3.9, 5.7) and 3.2 (2.5, 3.9), respectively. Of those with a record of attendance for biopsy it was possible to calculate PTC (baseline) HADS scores for 95%, baseline SF-12 scores for 96%, biopsy HADS scores for 79% and biopsy SF-12 scores for 79%. Complete ICSmale data were available at baseline for 95% of the 770 participants. The study group (Table 1) comprised the 569 (74%) of biopsy attenders with complete HADS, ICSmale (at baseline) and SF-12 scores.

Table 1.  Characteristics of PSA test recipients at baseline
VariablePSA test recipients (n)Study group*
AllAbnormal PSA result
  • *

    Men with complete HADS, ICS (at baseline) and SF-12 data who were later given an ‘abnormal’ PSA result and returned for biopsy

Age, years58.9 (7344)61.9 (855)61.9 (569)
Social class, %
I–II546059
III–IV343131
V–VI12 (7246)10 (810) 9 (547)
Urinary symptoms at baseline, %
Frequency27.0 (6917)30.7 (814)31.6 (569)
Urgency47.1 (6977)57.4 (815)57.8 (569)
Incontinence20.5 (6988)28.2 (815)29.5 (569)
Hesitancy46.1 (6962)57.6 (816)57.1 (569)
Nocturia17.1 (6957)26.1 (816)26.5 (569)
HADS at baseline, mean, 95% CI
Anxiety 5.2, 5.17–5.34 (6931) 4.83, 4.59– 5.07 (811) 4.87, 4.59–5.15 (569)
Depression 3.1, 2.99–3.12 (6933) 2.91, 2.74–3.10 (810) 2.90, 2.69–3.11 (569)
SF-12-MCS at baseline
MCS52.65, 52.44–52.87 (6960)53.25, 52.65–53.86 (816)53.21, 52.48–53.93 (569)

The mean age of the study group (those receiving an ‘abnormal’ PSA test result and attending for biopsy) was 61.9 years, with 63% aged >60 years; 59% were in social classes I-II and 31% in classes III–V. Men with an ‘abnormal’ PSA level who returned for biopsy were 3 years older than PSA test recipients generally, and more likely to report each of the five urinary symptoms at baseline than PSA test recipients generally, independently of age. Having one or more urinary symptoms was associated with an abnormal PSA level amongst test recipients, adjusting for age (P < 0.001). Biopsy attenders with missing data at biopsy had similar mean anxiety (4.70, 95% CI 4.15–5.24) and depression (2.85, 2.46–3.25) scores at baseline to those with complete data (Table 1).

Table 2 summarises the anxiety, depression and SF-12-MCS scores at baseline and at biopsy, and the change in those scores. The mean anxiety and depression scores were 4.9 and 2.9, respectively, at PTC, and 4.6 and 2.3 at biopsy. There was little change in scores when men returned for biopsy after receiving an ‘abnormal’ PSA test result. There was a slight reduction in the mean anxiety score (−0.27) and the mean depression score (−0.63).

Table 2. 
The depression (HADD), anxiety (HADA) and SF-12-MCS scores and the change in scores in the study group (569 subjects)
VariableBaselineBiopsyChange
HADA
 Mean (95% CI) 4.87 4.60−0.27 (−0.49, 0.05)
 sd 3.44 3.512.68
 Median 4 40
 10th-90th % 1, 9 0, 9−3, + 3
HADD
Mean (95% CI) 2.90 2.27−0.63 (−0.78, 0.48)
 sd 2.55 2.491.81
 Median 2 1−0.33
 10th-90th % 0, 6 0, 6−3, + 1
SF-12-MCS
 Mean (95% CI)56.3553.350.14 (−0.45, 0.74)
 sd 8.82 8.167.27
 Median56.3555.930
 10th-90th %40.72, 60.741.17, 60.25−8.67, 9.32

There was a positive association between HADS scores at baseline and at biopsy. The upper decile of the anxiety and depression scores had mean scores at baseline of 13.32 and 9.83, and 10.94 and 7.33 at biopsy, respectively, compared with overall mean scores at biopsy of 4.6 and 2.3. There was no evidence of a relationship between the initial score and change at biopsy (P = 0.49 for anxiety, P = 0.42 for depression).

There was a weak negative relationship between age and HADS scores in the study group. Age explained 2.0% (P < 0.001) of the variance in anxiety and 0.5% (P = 0.101) of the variance in depression at baseline. The mean anxiety score at baseline ranged from 5.7 for those aged 50–54 years to 4.2 for those aged 65–69 years (P = 0.003). The mean depression scores ranged from 3.3 for the youngest to 2.7 for the oldest group (P = 0.071). Change in depression and anxiety scores at biopsy were not associated with age.

Each of the five urinary symptoms was associated with higher levels of anxiety and depression at baseline and at biopsy (except frequency and nocturia with anxiety at biopsy, Table 3). The differences at baseline in anxiety scores between men who reported urinary symptoms and men who did not ranged from 0.9 for frequency to 1.7 for incontinence, and the differences in depression scores from 0.5 for frequency to 0.9 for incontinence (Table 3). The standardized difference in anxiety at baseline between men who did and did not report incontinence was the largest standardized difference in HADS scores between men who reported urinary symptoms and men who did not. This difference was 0.48 of a sd of the distribution of anxiety scores. However, 59% of participants reported several urinary symptoms, and 91% of men who reported incontinence also reported urgency. The difference in anxiety score between men reporting no (17% of participants) and five (7%) symptoms, adjusted for age, was 2.84 (0.83 of a sd).

Table 3.  The difference in HADS scores at baseline, HADS scores at biopsy and HADS scores at biopsy adjusted for baseline scores, between those who did and did not report urinary symptoms (adjusted for age at baseline) in the study group (569 subjects)
ScoreDifference in score at baselineDifference in score at biopsyDifference in score at biopsy adjusting for score at baseline
Anxiety scores (95% CI)
 Urgency1.48 (0.92, 2.04)0.95 (0.37, 1.53)−0.12 (−0.55, 0.31)
 Frequency0.9 (0.29, 1.49)0.54 (−0.07, 1.16)−0.10 (−0.54, 0.35)
 Nocturia0.73 (0.09, 1.37)0.56 (−0.11, 1.21)0.031 (−0.44, 0.51)
 Incontinence1.66 (1.06, 2.26)0.89 (0.26, 1.51)−0.32 (−0.78, 0.15)
 Hesitancy1.21 (0.65, 1.76)0.85 (0.27, 1.42)−0.02 (−0.44,0.41)
Depression scores (95% CI)
 Urgency0.59 (0.16, 1.02)0.64 (0.22, 1.05)0.21 (−0.07, 0.49)
 Frequency0.49 (0.04, 0.94)0.50 (0.06, 0.94)0.14 (−0.15,0.44)
 Nocturia0.54 (0.06, 1.03)0.75 (0.28, 1.22)0.36 (0.04, 0.68)
 Incontinence0.91 (0.45, 1.36)0.84 (0.40, 1.29)0.19 (−0.11,0.50)
 Hesitancy0.78 (0.35, 1.20)0.86 (0.45, 1.27)0.31 (0.03,0.59)

The relationship between these urinary symptoms and HADS scores adjusted for age was unaffected by also adjusting for SF-12-PCS score at baseline, i.e. poorer physical health as measured by the SF-12 did not account for the relationship between urinary symptoms and HADS scores (data not shown).

The relationship between HADS scores at biopsy and urinary symptoms was adjusted for HADS scores at baseline and age. All urinary symptoms together did not explain the variation in the adjusted scores (anxiety P = 0.85 and depression P = 0.11) in a multivariable model. There were no associations between any individual urinary symptoms and anxiety scores at biopsy in these models (Table 3). There was a weak relationship between depression at biopsy adjusted for baseline and both hesitancy (P = 0.039) and nocturia (P = 0.032; Table 3). However, given that all urinary symptoms together did not explain the variation in the depression score at biopsy (adjusted for baseline score), and as the differences are small (Table 3), these two apparent (weak) relationships with individual symptoms should be treated with caution. Adjusting these models for a measure of general health (SF-12-PCS) had no effect on the results (data not shown).

DISCUSSION

The receipt of an ‘abnormal’ PSA test result and attendance for further investigation did not appear to increase depression and anxiety as measured by HADS, or have a detrimental effect on general psychological health as measured by the SF-12-MCS, in men attending for biopsy. Urinary symptoms (except frequency and nocturia) reported before the PSA result were associated with higher levels of anxiety and depression at baseline and at biopsy, adjusting for age. However, when adjusting HADS scores at biopsy for baseline HADS scores, it appeared that the change in scores after receiving an abnormal PSA result was the same for men with and with no urinary symptoms. There was a positive relationship between HADS scores at baseline and at biopsy, men who had higher levels of anxiety and depression before the receipt of an ‘abnormal result’ also had higher scores at biopsy, but the psychological impact of receiving an abnormal PSA result did not appear to differ according to the baseline score.

It is possible that PSA screening might affect psychological health by increasing psychological distress in those attending for a PSA test, even if the receipt of an abnormal test itself has no further effect. These men did not appear to be much more anxious or depressed than would be expected of a population sample; levels of anxiety were the same as, and levels of depression less than, population norms for men in the UK not matched for age [16]. If the relationship found in the present study between age and anxiety (older men were less anxious) applied generally, the anxiety scores of men aged 50–69 years in the general population would be slightly lower than the population average, and the scores of men in this study possibly higher than expected. However, demographic variables (not including gender), have only very modest correlations with HADS in the general population [16,17]. Similarly, other studies of the psychological impact of PSA screening found that the psychological health of the men being screened at baseline was the same as, or better than, population norms [5,18].

The present findings concur with those of other European and USA studies that have used generic health status questionnaires. The Rotterdam trial [5], using the 40-item Spielberger state-trait anxiety inventory [19], and the EuroQol [20], found no statistically significant difference in anxiety or quality-of-life (QoL) scores on attendance at the screening centre (before the PSA test) and 2 weeks after biopsy (but before receipt of biopsy results). Essink-bot et al. [5] found that men with a high predisposition to anxiety had high levels of anxiety throughout the screening process. Even in men with high trait anxiety, anxiety did not increase after the receipt of an abnormal/suspicious screening test (PSA, DRE or TRUS) result [5]. State anxiety levels were higher immediately before screening and while awaiting the biopsy result than after a unsuspicious result; however, anxiety levels still did not differ much from the population norm (mean of 35 at biopsy vs 34; sd 11, for the Dutch general population). Similarly, a health-related (HR) QoL ancillary study to the Prostate, Lung, Colorectal and Ovarian Cancer (PLCO) screening trial, interviewed participants after screening test results but before biopsy and found no difference in SF-12 scores between those with normal and abnormal screening test results [18].

In the present study, there was a small improvement in psychological health, indicated by a decrease in mean HADS anxiety and depression scores, on attendance for biopsy. It is possible that men who were concerned that they might have prostate cancer would experience relief that something was being done. Alternatively, a decrease in scores might be due to a response shift (when people differently interpret and evaluate symptoms or QoL after the receipt of an ‘abnormal’ PSA result), or could be due to a practice effect (an increase in familiarity with the questionnaires at biopsy). There are some limitations to the general applicability of these data that should be addressed in future studies. It might be that this group differs from a general population-screening group because they elected to take part in a treatment trial (55% of men, aged 50–69 years and registered with a GP, responded to a written offer of a PSA test) and were consequently self-selected. However, qualitative research found that participants usually gave the assumed health benefits arising from screening and the reassurance provided by, and expectation of, a negative PSA test result as the main reasons for study participation [4]. The levels of urinary symptoms also appeared to be very similar to levels reported in a community sample in the UK. Men attending PTCs (baseline) reported similar levels of hesitancy (46%; 51%), urgency (47%; 48%), nocturia (18%; 14%), urge incontinence (21%; 20%) and frequency (27%; 33%), using the ICSmale questionnaire, to a population sample from a general practice [21]. Therefore, this group does not appear to be strongly self-selected based on urinary symptoms. The biopsy questionnaires were completed in NHS biopsy clinics and were collected for 81% of study participants. However, men with missing data at biopsy had similar mean depression and anxiety scores at baseline to other participants.

The authors of the Rotterdam study [5], reporting on the failure to detect any deleterious effect of screening, concluded that prostate cancer screening might be viewed by men as a routine examination and, as such, does not cause substantial psychological distress. Qualitative research in the ProtecT Study indicated that a few men appeared to express considerable concern after receiving an ‘abnormal’ PSA result, and most reported some level of concern [4], but this was not reflected in anxiety, depression or SF-12 scores in the present study or in the Rotterdam trial using a different measure of anxiety [5]. The study of QoL among participants in the PLCO screening trial [18], using items from the Intrusion Subscale of the Impact of Events Scale [22] to measure cancer-specific distress, found higher reported levels of intrusive thoughts about cancer amongst men who received an ‘abnormal’ than a normal screening test result, although there was no difference in HRQoL (SF-12) between the groups.

This pattern of findings might suggest that the measures of psychological health used were either not sensitive enough (and the effect on psychological health of any concerns about cancer was relatively minor) or that the instruments used do not adequately measure the psychological impact associated with receiving an abnormal screening test result. Hewitson et al. [23], in a review of studies exploring the impact of PSA screening and variables associated with psychological distress in men seeking and attending PSA screening, drew parallels with breast cancer screening, and argued that it was only with the development of a breast cancer screening-specific instrument, the Psychological Consequences Questionnaire [24], that ‘the subtle cognitive and emotional reactions’ to screening could be adequately measured [23]. Furthermore, the lack of association between urinary symptoms and the change in anxiety after an abnormal PSA result in the present study might be because generic measures of anxiety [25] are insensitive to cancer-specific worry.

In the present study, the HADS detected differences between men with and without urinary symptoms before and after an ‘abnormal’ PSA level result. To distinguish between anxiety associated with fear of a prostate cancer diagnosis and other causes of poorer psychological health in men with urinary symptoms, further investigation is needed into the relationship between urinary symptoms and the psychological impact of screening. The finding that poorer physical health does not appear to account for the relationship between urinary symptoms and HADS scores indicates that symptoms might increase psychological distress in men attending for screening, but it is also possible that men with higher levels of anxiety and depression are more likely to report symptoms. However, considering the ‘lay belief’ that urinary symptoms are associated with prostate cancer amongst men being investigated for symptoms in the UK [7], it might be beneficial to better inform those offered PSA testing about the uncertain relationship between urinary symptoms and prostate cancer and to highlight alternative explanations for common symptoms.

Previously published studies of the psychosocial impact of PSA screening in asymptomatic men tended to involve mostly men who had previous experience of prostate cancer screening and were not specific to a UK population [23]. The present study provided the opportunity to explore the impact of receiving an abnormal PSA result in a largely unselected UK population. We found that, although potentially raising participants’ concerns about prostate cancer, the receipt of an abnormal PSA test result among men at population risk of prostate cancer in the UK does not appear to have an impact on psychological health as measured by generic self-report questionnaires. Men reporting urinary symptoms were more likely to have higher levels of psychological distress before undergoing diagnostic and screening investigations than men with no symptoms, although again, these did not appear to be raised by the testing process per se. However, there might be a need to develop an instrument that is sensitive to specific concerns raised by PSA screening if the psychological implications of prostate cancer screening programmes are to be fully evaluated.

ACKNOWLEDGEMENTS

The ProtecT Study is funded by the Health Technology Assessment Programme (projects 96/20/06, 96/20/99). The authors thank Chris Metcalfe for his comments on earlier drafts of the paper and to acknowledge the tremendous contribution of all members of the ProtecT Study research group. We also thank all ProtecT Study respondents for their participation. Department of Health disclaimer: The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Department of Health.

CONFLICT OF INTEREST

None declared. Source of funding: Department of Health.

Ancillary