U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Hughes E, Mitchell N, Gascoyne S, et al. Sexual health promotion in people with severe mental illness: the RESPECT feasibility RCT. Southampton (UK): NIHR Journals Library; 2019 Dec. (Health Technology Assessment, No. 23.65.)

Cover of Sexual health promotion in people with severe mental illness: the RESPECT feasibility RCT

Sexual health promotion in people with severe mental illness: the RESPECT feasibility RCT.

Show details

Chapter 5Results

Statistical analysis

Baseline characteristics

Participant characteristics at baseline are shown in Table 5. The average age was 44.8 years, ranging from 22 years to 66.1 years. There was almost an equal split of men (48.6%) and women (47.2%); three participants stated that they were ‘other’, describing themselves as intersex, unable to select one answer, or not providing information. The majority of the participants classed themselves as heterosexual (81.9%), with four stating they were gay or lesbian (5.6%), six saying bisexual (8.3%), one participant preferring not to say and two responding ‘other’. Of the two that responded ‘other’ regarding their sexuality, one was unsure and one stated they were pansexual. Forty-six (63.9%) of the participants in the RESPECT study were white British; however, the other 36.1% of participants were from a wide range of ethnic backgrounds (see Table 5).

TABLE 5

TABLE 5

Baseline characteristics

Similarly, there was a wide range of religious beliefs represented in the trial participants. The most common was Christian (37.5%), or no religion (40.3%). However, the trial participants also included six Muslims (8.3%), one Sikh (1.4%), three Buddhists (4.2%), one participant preferring not to say (1.4%), and two (2.8%) stating that they had other beliefs.

The level of qualifications varied from six participants having no qualifications (8.3%), to seven participants having further higher degrees (9.7%). The most common level of education was AS (Advanced Subsidiary) level/A (Advanced) level and higher degrees, both having been completed by 12 participants (16.7%), with GCSE (General Certificate of Secondary Education)/GCE (General Certificate of Education)/CSE (Certificate of Secondary Education) having been obtained by 11 participants (15.3%). Over half of the participants in the trial were currently unable to work because of poor health (52.8%) and 13 were unemployed (18.1%). Of those that were currently in work, six were working full time and seven were working part time. Three participants (4.2%) were retired and three (4.2%) were students.

The majority of participants stated that they were living alone (58.3%), others were living with their parent/carers (15.3%), other relatives (5.6%), friends (4.6%) or their partner/spouse (8.3%). Three participants were living in hostels, and two had other living arrangements (one living in a house of multiple occupancy and the other in a supported living facility).

Primary outcomes

Recruitment rates

Over the course of 12 months, 138 people were screened. (This is based on data from the screening log. There is a larger proportion of people identified informally as eligible and packs were given to care co-ordinators or posted out; however, we were unable to get accurate information regarding how many of those potentially eligible people actually received the packs.) Of these 138 people, 117 people met the defined eligibility criteria for inclusion in the trial. This figure of 84.8% eligible people is much higher than the 50–60% anticipated.35,41 Of these, a total of 72 participants were randomised into the trial (March 2017 to January 2018), giving a recruitment rate of 61.5% (from which 52.2% of screened participants then went on to enter the study), again higher than the 40% which was predicted (Figure 4). The flow of participants can be seen in Figure 5.

FIGURE 4. Recruitment graph for the RESPECT trial from March to December 2017 (by centre).

FIGURE 4

Recruitment graph for the RESPECT trial from March to December 2017 (by centre).

FIGURE 5. The CONSORT flow diagram.

FIGURE 5

The CONSORT flow diagram. a, Reasons for not meeting inclusion criteria are not mutually exclusive; b, this eligibility criterion was dropped after 27 April 2017.

These 72 participants were recruited from the six sites, with LYPFT and C&IFT recruiting the majority (70.8%) of the participants, 24 and 27, respectively. The remaining 21 participants were recruited from SWYFT (n = 8), NELFT (n = 7), SPFT (n = 4) and the Community Links – aspire service (n = 2).

Retention rate

Intervention

Thirty-six participants (50%) were randomised to receive the intervention. The intervention consisted of three 1-hour sessions, each covering different material.

The first session was attended by 25 participants (69.4%), the second by 19 (52.8%) and the third by 18 (50%). In addition to this, five participants had combined sessions, in which they covered the material from multiple sessions at one time. In total, 17 participants attended all three sessions (47.2%), with 22 covering all the material from the three sessions (61.1%).

Twenty-five per cent (n = 9) of the participants never received any of the intervention sessions for various reasons; five had withdrawn from the intervention (see Withdrawals for more details) and reasons for the other four not attending included wanting a specific gender of interventionist (n = 1), logistical problems arranging appointments (n = 1) and other reasons (not known) (n = 2).

Completion of data

Questionnaires

The 3-month questionnaires were completed by 59 participants (81.9%); this was split equally across the two arms, with 30 from the intervention arm and 29 from the control arm. Owing to an extension to the recruitment period (during which there was only time to do the 3-month follow-up) not all participants reached the 6-month time point before the trial ended. Of the 72 that were randomised, 38 reached the 6-month time point (52.8%). Of these 38 participants, 29 completed the 6-month questionnaire (76.3%); these were split almost equally, 13 from the intervention arm and 16 from the control arm. Further details can be found in the CONSORT flow diagram (see Figure 5).

Completion of standardised measures

The details of the completion of the standardised measures can be found in Table 6. The percentages are given out of those who completed the relevant time point interview. These completion rates do not account for the attrition rates of the trial and, as such, represent the completion of the returned questionnaires. This is so that the standardised measures can be assessed for suitability for use as a future primary outcome, rather than assessing the level of attrition in the trial. For those measures in which there was no fixed scoring mechanism, or when not all questions were applicable to all, the completion rate is taken as a response to any element of the questionnaire.

TABLE 6

TABLE 6

Completion of the standardised measures used at each time point, as randomised

Overall, the standardised measures were all completed, with minimal missing data. There was also no visual difference in the completion rates between the treatment arms.

Identifying a possible future primary outcome

Although the completion rates of the various outcome measures that were implemented in the RESPECT trial were high, it was determined that none of these would be a suitable primary outcome. If a large-scale RCT were to be undertaken after the completion of this trial, the main focus would be around the proportion of unprotected sexual acts performed, which the intervention aims to reduce, and as such this would be the primary outcome. In the RESPECT trial, this was measured using the SERBAS; however, this measure also provided a lot of extra information that would not be deemed necessary in future and, as such, aspects of this measure may be taken or altered to be incorporated into a future trial, but the SERBAS as a whole would not be used.

In the RESPECT trial, a large proportion of the participants were not sexually active throughout the trial (n = 36; 50% reported no acts at any time point), which has reduced the sample on which the reduction of unprotected acts can be calculated. As such, there were only 22 (30.6%) participants who had been sexually active in the 3 months prior to baseline and at the 3-month follow-up: 11 intervention and 11 control participants (Table 7).

TABLE 7

TABLE 7

The number and percentage of total sexual acts undertaken without protection at baseline and at 3 months, and the change, by trial arm and overall

On average, the participants in the trial saw a reduction of 6.8 unprotected sexual acts over the 3 months between baseline and the 3-month follow-up, with a SD of 31.3 unprotected sexual acts. The control arm had an average increase of 2.3 in the number of unprotected sexual acts (SD 17.3) and the intervention arm had an average reduction of 15.9 unprotected sexual acts (SD 39.7).

On average, this equated to a 2.9% increase in the percentage of unprotected sexual acts; a 4.0% decrease for the intervention arm (SD 22.0) and a 9.7% increase (SD 32.2) for the control arm. It should be noted that of the 22 participants included, seven (31.8%) had no change in the proportion of unprotected sexual acts they undertook between baseline and 3 months. The 3-month time point was chosen as the point at which to measure this as all participants were still involved, whereas the 6-month follow-up was only reached by around 50% of the participants.

However, when looking at the 6-month time point, only nine participants have been sexually active both at baseline and at 6 months, and eight participants were in the control arm and one participant was in the intervention arm. There was an average 20.1% increase in the proportion of unprotected sex, with a SD of 33.5; however, it should be noted that this will be skewed by the intervention participant having a 100% increase in unprotected sexual acts, as the control arm had only a 10.1% increase.

For an external feasibility study, such as the RESPECT trial, to provide a robust estimate of the SD of an outcome measure to inform a sample size calculation, a minimum of 70 measured subjects is needed.45 Unfortunately, because of the recruitment problems and the large proportion of participants not participating in unprotected sexual acts in this instance, this had not been reached here.

Exploring possible clustering effects by therapist

As can be seen in Table 8, there were 11 different therapists involved in delivering the intervention to participants in the RESPECT trial. Of these 11, one therapist delivered 41.8% of the sessions, to 39.3% (n = 11) of the participants in the intervention arm. Five of the therapists delivered the intervention to only one participant each. As a result, the calculation of an ICC would be inappropriate at this time, as the data would not provide a reliable estimate.

TABLE 8

TABLE 8

Number of sessions conducted and number of participants treated by therapist

Possible future sample size

At baseline in the RESPECT trial, the participants undertaking any sexual acts were on average undertaking 70.1% of these without protection (SD 37.8). Using the percentage of unprotected sexual acts per person as the primary outcome for a future trial, a sample of 202 participants would be needed to power the trial to show a mean reduction of 15% in unprotected sexual acts from 70% in the control arm to 55% in the intervention arm, at 3 months after randomisation, with 80% power and 5% two-sided significance. Allowing for 25% attrition, as seen at the 6-month time point in the RESPECT trial, this increases this sample size to 270, with 135 participants per arm.

However, this number would need to be inflated further to account for those participants who may not be undertaking any sexual acts at 3 months but who were at baseline; in the RESPECT trial this was 6.8% of those who completed the 3-month assessment. Conservatively adjusting for this, using 10% loss, would increase the final sample size to 300 participants who currently engage in sexual activities.

Given that it took 12 months to recruit 72 participants into the RESPECT trial, from six sites, we would need to increase both the length of recruitment and the number of sites to reach a target of 300 participants. The six sites were open to recruitment for a combined total of approximately 40 months, and only 30 of the 72 recruited participants were sexually active when recruited; therefore, the recruitment in the RESPECT trial can be thought of as equivalent to 0.75 sexually active participants per site per month. In a future trial, allowing 18 months for the recruitment period would require 23 sites to be opened. Alternatively, 17 sites could be open for 24 months.

Alternatively, to detect a 10% reduction in unprotected sexual acts, with the same assumptions and conditions as above, a final sample size of 670 participants would be required. This would require 50 sites to be open for 18 months, or 38 sites for 24 months.

Secondary outcomes

As the RESPECT study is a feasibility trial and, as such, had a small sample size, no formal statistical comparisons have been conducted, and all comparisons are instead considered as purely indicative.

Sexual and risk behaviour measures

The SERBAS is a measure that is split into two separate questionnaires, one to be completed by males and one by females. Of the three participants who classified themselves as another gender, two completed the female version and one completed the male version. It is not known if the participant or the researcher selected which version to complete.

The main focus of the SERBAS in the RESPECT trial was quantifying the proportion of sex acts that were undertaken without any form of protection, with the hope that the intervention may influence this. Table 9 details the proportion of all sex acts (vaginal, oral and anal) that were undertaken without protection, within the last 3 months at each time point, overall, by treatment arm and by gender. Full details on the results from the SERBAS can be found in Tables 24 and 25 in Appendix 2.

TABLE 9

TABLE 9

The total number of sexual acts undertaken by gender, by study arm and overall, and the corresponding proportions of these acts that were unprotected

It can be seen in Table 9 that there is a high level of unprotected sexual acts occurring in the participants in the RESPECT study. However, it should be noted that there were participants in the study that were not sexually active over the course of the study. At baseline, only 20 participants who completed the female SERBAS (55.6%) and only 10 participants (27.8%) who completed the male SERBAS had undertaken any sexual act within the previous 3 months.

Within the RESPECT study there was almost no sex trading reported. At baseline, two participants reported having paid for sex once and that they had not used condoms on these occasions. At 3 months, a different participant reported having paid for sex once and did use a condom, and at 6 months one participant, who had also done so at baseline, reported that they had paid for sex three times and did not use a condom on any occasion. All of the participants who did not use a condom were in the control arm, and the participant who used a condom was in the intervention arm. Only one participant reported an episode of forced sex, which had occurred once. At the point of disclosure during the administration of the SERBAS (as per protocol) the researcher asked the participant if they had someone to discuss this with and they replied that they were discussing the incident with their care co-ordinator and it was being dealt with, and that no action or further support was required from the RESPECT study team. The chief investigator was informed of this and approved no further action.

From Tables 24 and 25 in Appendix 2, it can be noted that the female participants in the study reported having more vaginal sex than the men at baseline: an average of 37.3 times in the last 3 months for women, with 58.4% of that not involving condom use, compared with an average of five times for men, with 81.4% of that being unprotected. This gender difference continued at 3- and 6-month follow-up. It can also be seen that very few participants in the study reported having anal sex, but that almost all of the acts were undertaken without condoms. Similarly the majority of the oral sex (both participant on partner and vice versa) was reported as being undertaken without protection, despite a substantial amount of oral sex being reported. In this instance, protection could be either condoms or dental dams dependent on the gender of the participating people.

Participants were asked to select up to three reasons for not participating in sexual activity with male and with female partners in the last 3 months; not all participants utilised the possible six reasons. The majority of participants were not interested in having sex with a same-sex partner (n = 55) but the other reasons given (as frequencies) are as follows:

  • no current partner (n = 27)
  • participant not interested in sex (n = 10)
  • participant’s mental illness/medication side effects (n = 9)
  • not interested with opposite sex partner (n = 4)
  • participant’s illness or fatigue (n = 3)
  • lack of privacy (n = 3)
  • partner not interested in sex (n = 3)
  • partner died (n = 2)
  • partner’s mental illness/medication side effects (n = 1)
  • participant’s drug use (n = 1)
  • other reasons included: no opportunity (n = 2), does not trust women (n = 1), having to ‘fend off’ harassers (n = 1), religious reasons (n = 1), last partner cheated (n = 1), not looking for partner (n = 1), currently separated from wife (n = 1) and one participant gave no additional information.

Natsal-3

Elements of the Natsal-3 questionnaire were used within the RESPECT study. The results were similar across all three time points and the full details can be found in Table 26 in Appendix 2. One of the main areas covered by questions from the Natsal-3 questionnaire included contraception use. The male condom was found to be the most popular contraception method currently used, with 74.3% of men and 97.1% of women reporting using it as one of their most usual methods at baseline. Contraception was mainly obtained from health-care professionals (including doctors, or staff in pharmacies and sexual health clinics) and this was also found to be the favoured way to obtain contraception if participants had the option. Over half of the participants had visited a sexual health clinic, but most had not visited in the last year. Most participants had never been told that they had had a STI, and of those who had, chlamydia and genital warts were the most commonly reported STIs. Forty-two of the participants (58.3%) had been tested for HIV at some point, mainly as part of a sexual health check-up or because of concern for their own safety. For most of the participants this test was over 5 years ago, but eight participants (19.1%) had had a HIV test in the last year.

Knowledge about human immunodeficiency virus

The HIV-KQ consists of 18 questions, with a ‘yes’, ‘no’ and ‘don’t know’ option, meaning an average correct score can be calculated. Within the RESPECT study, only 17 questions were asked as it was felt that one of the questions, ‘A natural skin condom works better against HIV than does a latex condom’, was outdated and possibly confusing. The results are given in Table 10.

TABLE 10

TABLE 10

The HIV-KQ results, scored 0–100%, where higher scores represent greater knowledge

At baseline, the average score was 67.3% correct; this equates to between five and six incorrect answers on average. No one knew the answers to all of the questions and one person did not get any of the questions correct. The average score was similar in the two arms: 68.8% for the intervention arm and 65.9% for the control arm.

At 3 months, the average score had slightly increased (to 70.7%), equating to an average of five questions answered incorrectly. The average score was slightly different between arms: 75.7% for the intervention arm and 65.3% for the control arm. However, three participants now answered all questions correctly (two in the intervention arm and one in the control arm) with two participants answering all questions incorrectly, both from the control arm.

Despite the reduced number of participants who had data collected at 6 months, the results seen at 3 months still hold. The average score was 71.6% (74.3% for the intervention arm and 68.8% for the control arm). However, at this time point no participant received a score of 0.0%.

The most commonly incorrectly, or unknown, answered question was ‘All pregnant women infected with HIV will have babies born with AIDS’.

Motivations to engage in safer sex scale

The motivations to engage in safer sex scale consists of four questions asking how much risk the participant believes they are at of getting a STI, the chance of ever getting a STI, how big a problem STIs are in their community and whether or not they are worried about getting a STI. The majority of participants answered all questions in the ‘Somewhat at risk’ to ‘No risk at all’, with proportions staying constant across time points. One exception to this was the final question at month 3, which had an unexplainably large number of missing data (n = 33 of 59, 55.9%). Full details of the results can be found in Table 27 in Appendix 2.

Condom Use Self-Efficacy Scale

The Condom Use Self-Efficacy Scale is detailed in Table 11, and the results for sections one and two independently can be found in Table 28 in Appendix 2. The results between baseline and month 3 do indicate a slight rise in condom efficacy that drops slightly at month 6, but this may be explained by the reduced sample size. The Condom Use Self-Efficacy Scale was one of the least completed measures in the RESPECT trial (see Table 6 for completion rates); however, this is most likely to be because of the missing data mechanism used within this measure, in which only one item from each section could be missing and a score still calculated.

TABLE 11

TABLE 11

Self-efficacy scale results, scored 0 to 78, where a higher score represents greater condom use self-efficacy

Behavioural Intentions for Safer Sex

Scored from 0 to 40, the Behavioural Intentions for Safer Sex measure indicates the level of intention a participant has to have safe sex, with higher scores indicating more intention. Table 12 details the scores by arm and overall, at each time point. In Table 12 it can be seen that there is minimal change in the responses across the time points, but that it increases slightly for the intervention arm and that there is a slight decrease in the control arm.

TABLE 12

TABLE 12

Results of the Behavioural Intentions for Safer Sex measure, scored from 0 to 40, in which higher scores represent a greater intention to have safer sex

The Mental Illness Stigma Scale

Details of the results of the questions of the MISS-Q at each of the three time points can be found in Table 29 in Appendix 2. These results are given with the count and proportions who agreed with the statements, those who did not answer the question, and those who said that it was not applicable to them. One of the questions, which asked about protecting oneself or one’s partner from getting pregnant, has a larger than normal proportion of ‘N/A’ (not applicable) responses, which is because the question was printed incorrectly; it asked about protecting oneself from pregnancy and thus was not answered by most male respondents. At baseline, the statement with most agreement was ‘You are the one who chooses the course of your sexual life’ with 76.4% agreement overall. The statements with the least agreement were ‘Often a mental health provider has said you should not have a romantic or sexual relationship with people who do not have a mental illness’ and ‘Often a mental health provider has said you should have a hysterectomy, etc. as a form of birth control’, both with 0% agreement.

Alcohol, Smoking and Substance Involvement Screening Test

The ASSIST calculates risk scores for a set list of substances. The majority of the participants in the RESPECT trial were in the ‘Low risk’ category for all substances except tobacco, for which the majority of participants were categorised as at ‘Moderate risk’. These results were consistence across all three time points in the study. Table 13 details the total substance involvement scores, where the lower the score the less involvement the participant has had with any substance. The overall levels stay similar across the three time points. It can be seen that there are some participants who had no involvement with any substances (a score of zero) and some who had extensive involvement.

TABLE 13

TABLE 13

The ASSIST total substance involvement scores; higher scores represent a greater involvement with substances

Full details of the results of the ASSIST, including the breakdown of individual substances scores, can be found in Table 30 in Appendix 2.

Recovering Quality of Life

The ReQoL is a measure that looks at the quality of life of the participants, scored from 0 to 80, with a higher score indicating a higher quality of life. A score of < 50 is defined as characteristic of a clinical population. At each time point, the mean and median scores for the ReQoL were both < 50 (as detailed in Table 14), inferring that our population is typical of mental health service users. At baseline there were 42 participants (58.3%) with a score of < 50 (19 in the intervention arm and 23 in the control arm), at 3 months this was 36 participants (61.0%) (16 participants in the intervention arm and 20 participants in the control arm) and at 6 months it was 17 participants (58.6%) (eight participants in the intervention arm and nine in the control arm).

TABLE 14

TABLE 14

The ReQoL, scored from 0 to 80, where higher scores represent a better quality of life

Withdrawals

Through the course of the RESPECT study, 10 participants withdrew. Five participants withdrew from the intervention: one could no longer commit to the sessions as there had been a delay in receiving the intervention; one did not want to receive the intervention from a male and there was no other interventionist to deliver it; one withdrew after multiple cancelled sessions; and the other two gave no reason. Two participants were withdrawn from follow-up only, one after discussion with the chief investigator and clinician, and the other gave no reason. One of these participants was in the intervention arm and one was in the control arm. Full withdrawal from the trial was requested by four participants, with two in each treatment arm; one made a complaint to their local R&D office and three gave no reason. The complaint was made about the study focus (i.e. that it was about sex) after completion of the baseline interview, although the individual had read the information sheet and signed the consent form. The nature of the complaint was that we should not be undertaking research on the topic of sex. Our response was that we had NHS ethics approval to undertake the study and that the individual had had full information about the research topic during the recruitment process and was free to completely withdraw from the study, which they did. One of these full withdrawals had previously withdrawn from treatment; thus, there were a total of 11 instances of withdrawal.

Adverse events

No AEs were reported during the trial.

Health economics analyses

The objective of the economic analysis was to explore the feasibility of collecting cost-effectiveness data for a full RCT. As such, the economic analysis evaluated response rate, item completion rate and any emerging trends in the level and changes in health services resource use and health-related quality of life.

Health services resource use

Data completion

The questionnaire response rate was based on completion of at least one resource use item (including zeros). Table 15 shows that the number of respondents who completed at least one item of the resource use questionnaire was 36 (100%), 30 (83%) and 13 (36%) at baseline, 3 months and 6 months, respectively in the intervention arm. The response rate was comparable in the control arm with 36 (100%), 26 (72%) and 15 (42%) respondents completing the questionnaire at baseline, 3 months and 6 months, respectively. Moreover, this was similar to the overall questionnaire completion rate in the study.

TABLE 15

TABLE 15

Number of respondents with missing data, by treatment arm and time point

Table 15 also shows that the rate of missing items in the resource use questionnaire was very low. Response to an item was considered missing if response to the binary (yes/no) question about a service use item (such as GP consultations) was missing or the number of visits conditional on a ‘yes’ response was missing. In the intervention arm, one response was missing at baseline for GP consultations, A&E or urgent care visits and pharmacy visits. In the control arm, three responses were missing for pharmacy visits and one each for nurse and sexual health clinic consultations. Overall, there is a slightly higher number of missing values for pharmacy visits; however, overall the number of missing responses remained low.

Frequency and reasons for health service use

Table 16 shows the frequency of use of each type of health service in the last 3 months. Given the small number of respondents (particularly during the follow-up), the aim of this section is to summarise the results rather than draw a statistical inference of a difference between groups or patterns of service use over time.

TABLE 16

TABLE 16

Frequency of health service resource use in the last 3 months, reported at baseline, 3 months and 6 months in the intervention and control groups

At baseline, GP and nurse consultations were most common in both the intervention arm and the control arm, with 77.8% and 63.9% of respondents reporting at least one consultation with a GP or a nurse, respectively, in the intervention arm. The figures were similar in the control arm at 80.6% and 66.7% of respondents with a GP or nurse consultation, respectively. Moreover, the mean number of consultations for those who consulted was also similar in the two groups at baseline. During the follow-up period, the proportion of patients with GP or nurse consultations reduced over time at a similar rate in both the intervention and control groups, except at the 6-month time point when a higher proportion of respondents in the control arm (80%) had at least one nurse consultation than in the intervention arm (69%). However, given the small number of respondents at the 6-month time point, any statistically meaningful conclusion cannot be drawn.

Table 17 presents a summary of sexual health-related reasons for consultations for those who had at least one consultation (note: this may include more than one consultation by the same patient). For GP consultations, the two most common reasons for consultations at baseline were ‘to get advice for discharge through vagina, penis or anus; or sores in genital area; or pain when passing urine’ and ‘to get the morning-after pill or to discuss contraception or family planning’. For nurse consultations, a small number of visits were related to sexual health, with the two most common reasons being ‘to get tested for HIV, hepatitis or other STIs’ and ‘to discuss other sexual health issues not listed, including sexual assault’.

TABLE 17

TABLE 17

Reason for sexual health-related consultation for those who used a health service

Table 16 shows that, at baseline, a higher number (and proportion) of usual-care respondents (9 respondents, 25%) had at least one A&E or urgent care centre visit which was higher than the figure in the treatment arm (4 respondents, 11%). During the follow-up, the number of A&E (and urgent care) visits reduced over time in both arms. However, more importantly, none of these visits at baseline or at follow-up time points were related to a sexual health reason (see Table 17).

Pharmacy visits or appointments were few in both groups at baseline as well as at follow-up time points. The use of contraception clinics was low in both groups, with only three respondents using them in the intervention group at baseline, reducing to one and none at 3 and 6 months, respectively. Only one respondent in the control group used the service at 6 months. This, together with other responses related to accessing contraception, suggests that respondents were more likely to get contraception/family planning advice or the morning-after pill from a sexual health clinic or their GP than from a contraception clinic.

Sexual health consultations were relatively high at baseline in the intervention group (6 respondents, 16.7%) compared with the control arm (one respondent, 2.8%). These consultations reduced over time in the intervention group and remained low in the control arm. The most common reasons for consultations at a sexual health clinic were ‘to get tested for HIV, hepatitis or other sexually-transmitted infections’ and ‘to get the morning-after pill or to discuss contraception or family planning’. The use of the Sexual Assault Referral Centre (SARC) was also very low, with only one respondent each in the intervention and control arms using the service at baseline and no respondents using it at follow-up.

Unit costs of health service use

Unit costs were obtained from national databases, when available, and otherwise from other published sources. Sources of and assumptions for unit costs are presented in Table 18.

TABLE 18

TABLE 18

Unit costs of health services resource use

Intervention cost

The intervention consisted of up to three 1-hour sessions of sexual health promotion delivered by a trained mental health worker. These sessions delivered specifically designed training material. A total of 25 participants attended at least one session, 19 attended two sessions and 17 attended all three sessions (see Table 18). However, in some cases (n = 5) based on participants’ preference, a ‘combo’ session was delivered that covered material for more than one session, which reduced the number of sessions required/attended.

The duration of the three sessions was similar, with the mean being 59.2 minutes for the first session and 57.1 and 57.8 minutes for sessions 2 and 3, respectively (Table 19). This is in line with the time allocated for these sessions (i.e. up to 60 minutes per session). The duration of the combo sessions was longer with the mean being 93.8 minutes. There was some variability in the length of sessions, which is clear from the minimum and maximum values.

TABLE 19

TABLE 19

Time (in minutes) and cost of delivering the training sessions

Data were also collected on the proportion of the training material covered across all sessions by each participant who attended at least one session. This was used to estimate the total time required to deliver the full material to each participant; for example, if two-thirds of the material was covered in two sessions lasting 55 minutes each, then the estimated time to deliver the full material for this particular participant would be 165 minutes. Based on this approach, the estimated mean time to deliver the full training material was 170 minutes (range 101–307.5 minutes) (see Table 19). This was then multiplied by the hourly rate of a mental health worker (who delivered these sessions) which is £39 per hour (see Table 19). Based on this, the estimated mean cost of delivering the training material was £110.50 (range £65.70–199.90). However, it should be noted that the maximum value represents one outlier participant with one long (non-combo) session and excluding this participant changes the maximum cost to £146.

In addition to the cost of delivering the intervention, we considered the cost of training the interventionists (mostly, mental health workers). This included two components: (1) the time cost of the interventionist and (2) the time cost of the trainer. The training log (see Table 22) shows that the training was delivered over 1–2 days and each training session had between one and four participants. Overall, 21 people working in the mental health services received training to deliver the intervention. Initially it was delivered over 2 days but in the later stages it was condensed into 1 day. Overall, 16 training days were delivered by the study team. Assuming a full day equals 8 hours, and using the hourly wage of £39 for mental health workers, the estimated time cost for the interventionist to receive the training was £237.70. Regarding the second component of the training cost, we assumed that in practice the training will be delivered by a mental health worker (although during the project the training was delivered by the principal investigator of the research project). As a result, the total cost of delivering the training, including the time cost of the interventionist as well as the trainer, was equal to £475.20. Because this is a one-off cost, it was not added to the intervention delivery cost. Finally, although all the professionals providing the training were offered the option of using supervision support, none of them used this option.

Total cost of health service resource use

Figure 6 presents the total health services resource use cost in the last 3 months, by intervention arm and time point. This does not include the intervention cost. At baseline, resource use cost was slightly higher in the control arm than in the intervention arm: this is partly because of a higher number of A&E visits in the control arm than in the intervention arm. Health services costs reduced over time in both the intervention arm and the control arm, with costs in the control arm being higher than in the intervention arm at the 3-month time point and lower at the 6-month time point. These figures should be interpreted with caution because of the small number of respondents, particularly at the 6-month follow-up (at 6 months, n = 13 in the intervention arm and n = 15 in the control arm).

FIGURE 6. Health service resource use cost (excluding the intervention cost), by time point.

FIGURE 6

Health service resource use cost (excluding the intervention cost), by time point.

Health-related quality of life

Data completion

Table 20 shows the response rate for the EQ-5D questionnaire at baseline and at the 3-month and the 6-month follow-up. At baseline, all 36 respondents in the control arm completed all items in the questionnaire whereas one participant in the intervention arm did not complete the questionnaire. At the 3-month and the 6-month time points, EQ-5D completion rate was in line with the resource use questionnaire completion rate and the overall response rate (i.e. 30 and 13 responses at 3 months and 6 months, respectively, in the intervention arm, and 28 and 15 responses, respectively, in the control arm). Moreover, it is clear that, apart from those lost to follow-up, there were no missing data in EQ-5D items.

TABLE 20

TABLE 20

Data completion rate for EQ-5D items, by treatment arm and time point

Frequency and pattern of EQ-5D item responses

Figure 7 presents the distribution of completed responses to each domain or item of the EQ-5D questionnaire at baseline and at follow-up in the intervention arm. For mobility and self-care domains, over 70% of respondents in the intervention arm had no problem or slight problems at baseline and follow-up. However, there was a small shift in mobility from moderate problems at baseline and 3 months to improvement (slight problems) or worsening (severe problems) at 6 months. For the self-care domain, there was a small improvement over the follow-up period in the intervention arm, which was observed as a reduction in severe problems. For the usual activities domain, just over 50% of respondents in the intervention arm reported having some level of problems. There was a small improvement in usual activities at 6 months with no respondent reporting severe or extreme problems. For the pain and/or discomfort item, just under half of all respondents in the intervention arm reported having some level of problems at baseline. Over the follow-up period, the proportion of respondents with severe problems reduced but at the same time there were fewer respondents with no problems. Finally, for the anxiety and/or depression domain, most respondents in the intervention arm reported having some level of problems; however, there was a small improvement during the follow-up in terms of a reduction in respondents with severe or extreme problems.

FIGURE 7. The EQ-5D responses at baseline (BL), 3 months (3M) and 6 months (6M) for the intervention group.

FIGURE 7

The EQ-5D responses at baseline (BL), 3 months (3M) and 6 months (6M) for the intervention group.

Figure 8 presents the distribution of completed responses to EQ-5D domains at baseline and follow-up in the control arm. Approximately 90% and 80% respondents in the control arm had no problem or slight problems at baseline on the mobility and self-care domains, respectively. This suggests that the control arm was slightly healthier at baseline than the intervention arm. The same was true for the usual activities, pain/discomfort and anxiety/depression domains at baseline. During the follow-up period, the mobility, self-care and pain/discomfort domains had minimal change but there was an increase in slight problems in the usual-care activities domain at 6 months compared with baseline. Finally, there was an increase in the percentage of respondents with no problems as well as those with moderate problems in terms of anxiety and/or depression.

FIGURE 8. The EQ-5D responses at baseline (BL), 3 months (3M) and 6 months (6M) for the usual-care group.

FIGURE 8

The EQ-5D responses at baseline (BL), 3 months (3M) and 6 months (6M) for the usual-care group.

Health-related quality of life (utility values)

Figure 9 compares utility values for the 22 patients who completed three sessions of the intervention with the control arm. The figure shows greater improvement in utility value in the intervention arm than for usual care, and also than for the overall intervention group. However, as with the cost data, these results should be interpreted with caution as they represent only the complete cases and a small sample size. Moreover, the intervention group in Figure 9 represents a selected group who underwent the maximum number of sessions and may not represent the overall target population. This is in line with Figures 7 and 8, which also showed a slightly bigger improvement in quality-of-life domains in the intervention group than in the control group.

FIGURE 9. Utility values based on EQ-5D-5L responses at baseline, 3 months and 6 months for the treatment (those who completed three sessions) and usual-care groups.

FIGURE 9

Utility values based on EQ-5D-5L responses at baseline, 3 months and 6 months for the treatment (those who completed three sessions) and usual-care groups.

Copyright © Queen’s Printer and Controller of HMSO 2019. This work was produced by Hughes et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.
Bookshelf ID: NBK551246

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.0M)

Other titles in this collection

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...