Suicide, self-harm and suicidal ideation have long been associated with HIV infection [1–3]. Prior to treatment availability, rates of suicidal behaviours were seen as high in all groups studied [4,5]. However, it was unclear whether risk factors for suicidal behaviour were associated with those groups more vulnerable to HIV infection in the first place, and to what extent an HIV diagnosis contributed to this psychological trauma [6,7]. Attempted suicide is a strong predictor of emotional distress and completed suicide [8–10]. Suicidal ideation is well documented as a state that may precede suicide planning and attempted suicide  and is studied as a mental health burden in its own right as well as an indicator for future mental health burden .
Suicidal ideation is a distressing psychological phenomenon, indicative of low mood and poor quality of life. It has been well established that gay men , drug users  and those with diagnosed mental illness  are at heightened risk for suicidal ideation, attempts and completions. Given that these groups are disproportionately affected by HIV infection, it seems reasonable to assume that emotional support and specific HIV-related care should bear in mind their burdens contributing to self-harm [3,15]. It is well established that those who complete suicidal acts have a high chance of a prior history of suicidal ideation . Suicide prevention studies point to the importance of monitoring suicidal ideation  and responding appropriately to it as a key suicide prevention strategy .
Although the literature prior to new treatments shows a significant association between HIV infection and self-harm, very little has been systematically examined since treatment availability [7,19,20]. It may be reasonable to assume that treatment availability and the improved prognosis of HIV infection may dramatically affect suicidal ideation, attempts and completions, this has not been systematically investigated. From 1990 to 2002, 271 HIV-positive people committed suicide in the United Kingdom. The odds ratio of suicide as a cause of death for HIV-positive people was recorded as 0.9 for the period 1990–1994 and 5.3 for the period 1997–2002– the first era of treatment availability. New treatments have dramatically changed the nature and prognosis of HIV infection. The prognosis for those accessing treatment has been transformed . Studies have tried to document the various components and rates of optimism surrounding new treatments and have noted that such optimism is higher among the HIV positive than the HIV negative , and the treatments themselves bring with them a balance between improved prognosis, restored health and a burden of lifelong adherence needs, possible challenging side-effects and the roller coaster of treatment failure, treatment switching and treatment stopping . Furthermore, the data clearly demonstrate that optimism does not prevail , and a sense of realism is a more accurate description of the mood of HIV-positive respondents who are asked about treatment and infectiousness optimism in the light of new treatments.
Suicidal ideation prevalence among community samples of gay men appears to have remained stable over time (ranging from 13.1% in 2001 to 12.7% in 2004) . Yet, there is no literature comparing HIV-positive heterosexual women, heterosexual men or recent data on gay men with HIV infection in the era of treatment. This study aimed to measure the 7-day prevalence of suicidal ideation among HIV-positive clinic attenders in London and the Southeast and test the relative contribution of factors associated with such ideation in order to enhance mental health services for people with HIV infection in the era of antiretroviral therapy.
Data were collected from consecutive attenders at five HIV outpatients clinics during a 3-month period in 2005/2006. Following ethical approval, all consecutive attenders were assessed for inclusion in the study. Of the 1006 attenders over the time period, 903 met the eligibility criteria which required sufficient knowledge of English to complete the questionnaire and to provide informed consent to participate. All participants provided written consent. Anonymous questionnaires were completed and returned either via a clinic-based post box or a reply-paid postal system according to respondent preferences. Questionnaires were constructed to examine the following.
- Suicidal ideation: This was measured using a self-report item, based on feelings in the preceding week. Respondents were asked to indicate whether they had suicidal thoughts over the preceding week (occurrence) and were further asked to rate the frequency of such thoughts if they occurred (intensity) on a four-item scale (Rarely, Occasionally, Frequently, Almost Constantly). The scale was based on the Memorial Symptom Assessment Schedule Short-form (MSAS-Short-form) inventory coding system, and the suicidal ideation question was informed by the methodology used by Warner et al. and framed according to the MSAS-Short-form inventory . The MSAS-Short-form has good psychometric properties. Those recording suicidal ideation were compared with those not recording suicidal ideation.
- Demographic information was gathered on standardized forms to include age, sex, employment, ethnicity, residency, sexuality and education. Sexuality was self-reported (gay/bisexual/heterosexual/other).
- Relationship and sexual behaviour in the preceding 3 months was monitored to provide information on stable relationships, sexually transmitted infections and unprotected sexual intercourse with partners of unknown or different HIV status. Risky sexual behaviour was operationally defined as unprotected sex (no condoms) with a partner of unknown or different HIV status in the preceding 3 months.
- HIV optimism: Two measures of optimism were scored on a five-point Lickert scale. These were levels of optimism in relation to treatment and infectiousness on a standardized scale .
- Status disclosure levels were recorded based on questions from Kalichman et al.'s study  which provided details of any disclosure and disclosure details in case there had been disclosure to others.
- Pain and symptoms were measured using the MSAS-Short-form standardized inventory reporting the multiple item prevalence and burden for the previous 7 days, including three subscales of physical symptoms, psychological symptoms and Global Distress Index . This is a well validated scale with good psychometric properties and used in large-scale studies of HIV as well as other disease conditions such as cancer.
- Health-related quality of life was measured using the EuroQol measures giving two scales: the EuroQol-5D  comprises five elements– mobility, self-care, activities, pain/discomfort and mood; the EuroQol-visual analogue scale (VAS) generates a rating of overall general health supplemented by a self-rating score on quality of life. The EuroQol is a well validated measure used within HIV-related studies as well as more generally. Quality of life measures that are standardized and adapted to HIV-specific issues are still evolving.
- Adherence: Data on adherence used self-reported adherence over the preceding week. Ninety-five percent adherence rate was utilized as dose ‘adherent’. Further measures were taken of dose timing and adherence to circumstance requirements (most notably food and fasting). Full adherence was categorized as a combination of dose adherence, dose timing and circumstance adherence. Partial adherence was categorized as dose adherence above 95% but not full timing and circumstance adherence. Nonadherence was categorized for all remaining participants.
- Treatment: Details of antiretroviral treatment status, number of treatment switches and reasons for switching were taken.
- Clinic note data: For a subset of patients who gave specific informed consent, additional information from the clinic database was extracted and matched with questionnaire data. Additional items included recorded route of HIV transmission, time since HIV diagnosis, CD4 cell count, viral load and type of highly active antiretroviral therapy (HAART) regimen [protease inhibitor or nonnucleoside reverse transcriptase inhibitor (NNRTI)-based] at the time of questionnaire completion.
All data were entered into SPSS v14 (SPSS Inc., Chicago, Illinois, USA) for analysis. Descriptive statistics were used to describe sample characteristics. The association of suicidal ideation with demographic, behavioural, symptom and treatment variables was assessed, using the χ2 test for categorical variables and the two-sample t-test for continuous variables.
In order to identify variables independently associated with suicidal ideation, a multivariable logistic regression model was used. Covariates were entered into the model if a significant association (P < 0.05) was found in univariate analysis. Variables with P less than 0.1 are presented in the final multivariable model. Predictors of suicidal ideation were also assessed among the subgroup of patients taking antiretroviral treatment at the time of the questionnaire.
During the study period, 1006 people attended the five clinics and 903 met the eligibility criteria for inclusion. Of these, 778 agreed to participate and completed a questionnaire giving an overall response rate of 77% of all patients and 86% of eligible patients. The patient group comprised 24.2% (n = 183) heterosexual women, 10.1% (n = 76) heterosexual men and 65.7% (n = 496) gay/bisexual men. The group consisted of 51.3% (n = 396) UK-born participants and 48.7% (n = 376) non-UK born participants; 77.5% of respondents had been resident in the United Kingdom for more than 5 years whereas 22.5% had resided for less than 5 years. The majority (513; 67.3%) self-coded their ethnicity as White, with 188 (24.7%) coding as Black and 61 (8.0%) as Asian and Mixed/Other. The majority (377; 53.0%) were employed. University level education was reported by 328 (43.3%) respondents– which is high for a community sample, but in line with other reports on HIV clinic samples. Treatment experience showed 155 (20.8%) as treatment naive and 623 (79.2%) having taken antiretroviral therapy (ART)– 161 (21.6%) were on their first treatment, 135 (18.1%) had one treatment switch, 196 (26.3%) had multiple switches and 99 (13.3%) had stopped taking their treatment.
Rates of suicidal ideation
Suicidal ideation rates in the preceding week for the sample were high, with 31% (n = 245) of the group reporting suicidal ideation in the previous 7 days. Of the 227 who gave information on the frequency of suicidal thoughts, 12 (1.5% of the total sample, 5.3% of those with ideation) said this was constant, 24 (3.1% of the total, 10.6% of those with ideation) said this was frequent and the remainder reported an occasional/rare intensity.
Impact of suicidal ideation
Univariate comparisons were conducted (see Table 1) on demographic, treatment and behavioural variables according to presence or absence of suicidal ideation in the preceding week (irrespective of intensity). Heterosexual men were significantly more likely than women and gay men to express suicidal ideation (44.7 vs. 31.1% and 30.0%, respectively, χ2 = 6.6, P = 0.04). There was a trend for non-UK born respondents to have higher suicidal ideation (34.6 vs. 28.5%, χ2 = 3.3, P = 0.07) but no effect according to length of residency in the UK. Those who reported black ethnicity had elevated suicidal ideation rates (39.4 vs. 29.6% Whites and 26.2% Asian and Other, χ2 = 7.0, P = 0.03). There was a trend for those with university education to have lower suicidal ideation rates (28.0 vs. 33.9%, χ2 = 2.9, P = 0.09). Suicidal ideation rates were significantly lower among the employed (23.9 vs. 38.9% unemployed, χ2 = 18.8, P < 0.001), among those who had disclosed their status to someone else compared with those who had not (30.7 vs. 48.9%, χ2 = 6.8, P = 0.009) and those in a stable relationship compared with those who were not (27.4 vs. 37.0%, χ2 = 7.5, P = 0.006). There were no differences according to recent sexually transmitted infection (STI) diagnosis, number of sexual partners in the preceding 3 months, risky sex or optimism ratings around treatment and infectiousness. Patients who had stopped taking their treatment had higher suicidal ideation rates than those who were on treatment or were treatment naive (χ2 = 10.0, P = 0.04).
Among the subgroup of 492 patients on treatment (Table 2), suicidal ideation was significantly higher for those who were classified as nonadherent (43.1%) compared with partially adherent (26.7%) and fully adherent (26.4%) (χ2 = 10.5, P = 0.005).
There was no significant association between suicidal ideation and age (see Table 3). Those who expressed suicidal ideation had significantly higher scores on physical symptom burden (1.13 vs. 0.62, t = 10.4, P < 0.001), psychological burden (1.9 vs. 1.02, t = 12.5, P < 0.001) and global burden (1.61 vs. 0.89, t = 12.5, P < 0.001) on the MSAS scales. Furthermore, quality of life was rated significantly lower for those who expressed suicidal ideation on both measures of the (EuroQol) quality of life. For the 331 patients who had switched their treatment regimen, significantly more reasons were given for the subgroup reporting suicidal ideation than those not reporting suicidal ideation (mean 2.1 vs. 1.3, P = 0.001 for major triggers; 1.2 vs. 0.7, P = 0.002 for minor triggers).
Multivariate analysis: covariates significantly associated with suicidal ideation
The following variables, significant at the 5% level in invariable analysis, were entered into a logistic regression model with suicidal ideation as the dependent variable: sex/sexuality group (three levels), ethnicity (three levels), employment (two levels), disclosure of HIV status (two levels), whether been in a stale relationship (two levels), treatment switching group (four levels), MSAS-physical, MSAS-psychological and EuroQol score (these last three as continuous variables). Table 4 shows the independent associations with suicidal ideation; results are expressed as odds ratios with 95% confidence intervals (CIs). Sex/sexuality and ethnicity were independently associated with suicidal ideation: the odds of suicidal ideation were increased almost two-fold for heterosexual men compared with gay men or to women and for black patients compared with White or Asian patients. Lack of disclosure of HIV status was independently associated with a two-fold increase in odds of suicidal ideation. The MSAS-physical and MSAS-psychological scores were strongly and independently associated with suicidal ideation. Employment, whether they had been in a stable relationship, treatment group and EuroQol score were not significant in the multivariable model.
An additional model considered the subgroup of 492 patients on antiretroviral treatment only, of whom 148 (30.1%) reported suicidal ideation. The independent associations of demographic, behavioural and symptom factors with suicidal ideation were very similar to those seen in the sample as a whole (Table 5). Switching status and nonadherence were not significantly independently associated with suicidal ideation among patients on treatment.
Suicidal ideation, HIV transmission group, time since diagnosis, CD4 cell count and viral load
Of all 788 patients, 526 (67.6%) gave consent for the questionnaire information to be matched with information from their clinic notes. Information was matched in 478 cases; the remainder provided an incomplete hospital number. The proportion of patients who gave consent was similar among those with (66.9%) and without (67.9%) suicidal ideation. Suicidal ideation was much more common among the small number of patients with injection drug user (IDU) transmission group (seven out of 12; 58.3%) than those with heterosexual (48 out of 137; 35.0%) or homosexual (92 out of 326; 28.2%) transmission (χ2 = 6.4, P = 0.04). The proportion of patients reporting suicidal ideation did not vary significantly according to time since diagnosis [20 out of 50 (40.0%), 18 out of 79 (22.8%), 111 out of 349 (31.8%) for ≤1 year, >1–3 years, >3 years since diagnosis, respectively, χ2 = 4.5, P = 0.11]. Patients with CD4 cell count 200/μl or less tended to be more likely to report suicidal ideation (21 out of 50; 42.0%) than those with CD4 cell count more than 200/μl (125 out of 420; 29.8%; χ2=3.1, P = 0.077). The proportion of patients with suicidal ideation was similar in those with viral load 50 cells/ml or less (87 out of 280; 31.1%) and those with viral load more than 50 cells/ml (53 out of 175; 30.3%; χ2 = 0.03, P = 0.86). In the subgroup of 310 patients taking antiretroviral treatment, suicidal ideation was not significantly associated with undetectable viral load, low CD4 cell count, use of a protease inhibitor or use of an NNRTI (data not shown).
The present study reveals an alarming 7-day prevalence of suicidal ideation among HIV-positive clinic attenders, despite the current enormous advances in treatment and prognosis. This cannot be accounted for by the preexisting and established elevated suicidal ideation rates among gay men, as the rates were consistent among HIV-positive women and were specifically elevated among HIV-positive heterosexual men. Nor can it be accounted for by those receiving hepatitis C virus (HCV) therapy, given that the number of drug users within the sample was very low (n = 12). Special attention seems to be needed for heterosexual men who exhibited the highest prevalence. The rates may well be compounded by ethnicity (most heterosexual men were also non-UK born). The particular issues of stigma and shame as well as immigration status and insecurity may all be potential contributors to mood fluctuations. Suicidal rates were lowest among the employed and highest among the long-term sick. These data show similarities with the general suicidal literature on the impact of long-term illness and illness burden on all aspects of suicidal behaviour.
Stable relationships were protective for suicidal ideation. Unlike previous literature which has associated poor mental health with elevated sexual risk, no such association was found in this study. Although this is advantageous for HIV prevention issues, the same was not found in relation to treatment adherence. For those on treatment, respondents who were noted as nonadherent in the preceding week had significantly higher suicidal ideation rates than the partially or fully adherent group. Interventions to alleviate thoughts of suicidal ideation may improve adherence. However, these associations were not independent of demographic and symptom factors.
Particular attention needs to be given to those people who have stopped taking treatment. The group of treatment stoppers (n = 99, 12% of total sample) had heightened suicidal ideation rates (43.4% of all stoppers). This may be related to their lack of hope in treatment or their inability to benefit from antiretroviral treatment.
Although only 6% of the sample had not disclosed their HIV status to anyone, this group had a disproportionate prevalence of suicidal ideation (nearly one in two of this group reported ideation in the preceding week), and nondisclosure was an independent predictor of suicidal ideation. The burden of secrecy and the lack of community and social support appears to weigh heavily on the mental health of people with HIV infection who feel unable to disclose their status. Support and assistance with disclosure, social contact and emotional support are priorities for this group.
The symptom burden for everyday discomforts has been noted as high , among HIV-positive people on treatment. These data show that those with elevated physical and psychological burden scores had significantly higher suicidal ideation rates. Not surprisingly, those who expressed suicidal ideation scored lower on measures of quality of life. The regression data show that physical and psychological burden scores were the most notable independent predictors of suicidal ideation. Our data suggest that clinics need to pay particular attention to the burden of symptoms– both physical and mental– and not to focus purely on treatment regimens and outcome indicators of successful viral suppression.
Lastly, it is notable that achieving viral suppression to undetectable levels, time since diagnosis, use of a protease inhibitor or use of an NNRTI was not associated with likelihood of suicidal ideation.
The study has a number of limitations which would suggest caution in interpreting these findings. Although the response rate is high, and the coverage of HIV clinic attenders robust, the cross-sectional nature of the data and the limited measurement of suicidal ideation do not provide detailed information on causes of such ideation and the course of ideation over time. Future research is necessary to provide longitudinal data. The data clearly indicate the importance of suicidal ideation monitoring and the silent burden carried by many HIV-positive clinic attenders.
Despite these limitations, the data clearly underscore the importance of mental health burden and demonstrate the disturbing prevalence of suicidal thoughts, that is nearly a third of HIV-positive clinic attenders. Clearly, these data indicate the urgent need to address the underlying predictors of such ideation in treatment plans and highlight the ongoing need for mental health services despite the availability of life-prolonging treatments. Furthermore, some of the traditional mental health triggers of suicidal ideation may need rethinking in the era of HIV, where a host of other infection-related burdens, perhaps not well addressed by mental health services, may contribute to suicidal ideation, such as immigration law, access to services and structural violence. As suicidal acts are still notable among causes of death for HIV-positive people , these data suggest that full attention to suicide prevention initiatives should be incorporated into routine care.
We wish to acknowledge the contribution of Sally Norwood, Heather Leake-Date, Ms A Jayakody, research nurses at participating clinics and all survey respondents. This research was assisted with an unrestricted educational grant from GlaxoSmithKline, with input from the Adherence Strategy Group. L.S. was principal investigator for the study and formulated the design, tools analysis and implementation. She took primary responsibility for writing the article. R.H., joint principal investigator, contributed at all stages and assisted in report drafting and comments. F.L. was lead statistician and provided input on drafting, analysis and wrote sections of the article. M.F., J.A., M.J., S.E., G.A. HLD and S.Z. facilitated access to clinics, commented on questionnaires and commented on drafts of the article.
1. Perry S, Jacobsberg L, Fishman B. Suicidal ideation
and HIV testing. JAMA 1990; 263:679–682.
2. Sherr L. Suicide and AIDS: lessons from a case note audit in London. AIDS Care 1995; 7(Suppl 2):S109–S116.
3. Starace F, Sherr L. Suicidal behaviours, euthanasia and AIDS. AIDS 1998; 12:339–347.
4. King M, McKeown E, Warner J, Ramsay A, Johnson K, Cort C, et al
. Mental health and quality of life of gay men and lesbians in England and Wales: controlled, cross-sectional study. Br J Psychiatry 2003; 183:552–558.
5. Sherr L, Barnes J, Elford J, Olaitan A, Miller R, Johnson M. Women with HIV disease attending a London clinic. Genitourin Med 1997; 73:274–279.
6. Abelson J, Lambevski S, Crawford J, Bartos M, Kippax S. Factors associated with ‘feeling suicidal’: the role of sexual identity. J Homosex 2006; 51:59–80.
7. Carrico AW, Johnson MO, Morin SF, Remien RH, Charlebois ED, Steward WT, et al
. Correlates of suicidal ideation
among HIV-positive persons. AIDS 2007; 21:1199–1203.
8. Harris EC, Barraclough B. Suicide as an outcome for mental disorders. Br J Psychiatry 1997; 170:205–228.
9. Kessler RC, Borges G, Walters EE. Prevalence of and risk factors for lifetime suicide attempts in the National Comorbidity survey. Arch Gen Psychiatry 1999; 56:617–626.
10. Neeleman J, Wessely S, Wadsworth M. Predictors of suicide, accidental death, and premature natural death in a general-population birth cohort. Lancet 1998; 351:93–97.
11. Hawton K, Rodham K, Evans E, Weatherall R. Deliberate self harm in adolescents: self report survey in schools in England. BMJ 2002; 325:1207–1211.
12. Warner J, McKeown E, Griffin M, Johnson K, Ramsay A, Cort C, et al
. Rates and predictors of mental illness in gay men, lesbians and bisexual men and women: results from a survey based in England and Wales. Br J Psychiatry 2004; 185:479–485.
13. Lloyd JJ, Ricketts EP, Havens JR, Cornelius LJ, Bishai D, Huettner S, et al
. The relationship between lifetime abuse and suicidal ideation
in a sample of injection drug users. J Psychoactive Drugs 2007; 39:159–166.
14. Haw C, Hawton K, Sutton L, Sinclair J, Deeks J. Schizophrenia and deliberate self-harm: a systematic review of risk factors. Suicide Life Threat Behav 2005; 35:50–62.
15. Kalichman SC, Heckman T, Kochman A, Sikkema K, Bergholte J. Depression and thoughts of suicide among middle-aged and older persons living with HIV-AIDS. Psychiatr Serv 2000; 51:903–907.
16. Zahl DL, Hawton K. Repetition of deliberate self-harm and subsequent suicide risk: long-term follow-up study of 11,583 patients. Br J Psychiatry 2004; 185:70–75.
17. Hawton K, Fagg J. Suicide, and other causes of death, following attempted suicide. Br J Psychiatry 1988; 152:359–366.
18. van der SR, van RL, Buskens E, Allart E, Hawton K, van der GY, et al
. Intensive in-patient and community intervention versus routine care after attempted suicide. A randomised controlled intervention study. Br J Psychiatry 1997; 171:35–41.
19. Shelton AJ, Atkinson J, Risser JM, McCurdy SA, Useche B, Padgett PM. The prevalence of suicidal behaviours in a group of HIV-positive men. AIDS Care 2006; 18:574–576.
20. Sherr L, Bolding G, Davis M, Hart G, Elford J. Suicide concerns and HIV among London gay men in the post HAART era
. In: The XV International AIDS Conference, Bangkok, 2004 [abstract MoOrD1030].
21. Nylen G, Mortimer J, Evans B, Gill N. Mortality in young adults in England and Wales: the impact of the HIV epidemic. AIDS 1999; 13:1535–1541.
22. Egger M, May M, Chene G, Phillips AN, Ledergerber B, Dabis F, et al
. Prognosis of HIV-1 patients starting highly active antiretroviral therapy: a collaborative analysis of prospective studies. Lancet 2002; 360:119–129.
23. Elford J, Bolding G, Sherr L. High-risk sexual behaviour increases among London gay men between 1998 and 2001: what is the role of HIV optimism? AIDS 2002; 16:1537–1544.
24. Sherr L, Lampe F, Norwood S, Leake-Date H, Fisher M, Edwards S, et al
. Successive switching of antiretroviral therapy is associated with high psychological and physical burden. Int J STD AIDS 2007; 18:700–704.
25. International Collaboration on HIV Optimism. HIV treatments optimism among gay men: an international perspective
. J Acquir Immune Defic Syndr
26. Chang VT, Hwang SS, Feuerman M, Kasimis BS, Thaler HT. The memorial symptom assessment scale short form (MSAS-SF). Cancer 2000; 89:1162–1171.
27. Kalichman SC, DiMarco M, Austin J, Luke W, DiFonzo K. Stress, social support, and HIV-status disclosure to family and friends among HIV-positive men and women. J Behav Med 2003; 26:315–332.
28. Chang VT, Hwang SS, Kasimis B, Thaler HT. Shorter symptom assessment instruments: the Condensed Memorial Symptom Assessment Scale (CMSAS). Cancer Invest 2004; 22:526–536.
29. The EuroQol Group. EuroQol: a new facility for the measurement of health-related quality of life. The EuroQol Group
. Health Policy
30. Harding R, Molloy T, Easterbrook P, Frame K, Higginson IJ. Is antiretroviral therapy associated with symptom prevalence and burden? Int J STD AIDS 2006; 17:400–405.
31. Health Protection Agency, CDR weekly reports, accessed at http://www.hpa.org.uk/cdr/archives/1994/cdr3894.pdf