Our results suggest that in England around 96% (n = 6857) of adults aged 16 or over have had at least one partner or spouse. Men (5.4%, n = 122) were more likely than women (2.7%, n = 79) to have never had a partner (p < 0.001). Subsequent analyses are based on participants reporting at least one partner in their life, to avoid results being confounded by relationship history.

One person in fifteen (6.6%; n = 484; 95% CI 5.9–7.3) had received two or more unwanted obscene or threatening texts, emails, letters, or cards, that had caused them fear, alarm or distress from a current or former partner. This was twice as likely in women (8.7%) as men (4.4%) (p < 0.001). 14.5% of women aged 16–24 reported receipt of threatening/obscene messages, three times the rate for men of the same age (4.3%). A quarter (25.1%) of people exposed to threatening/obscene messages had received more than one in the past year (1.7% of adults, n = 124). Two-fifths (40.1%) of those with past-year exposure had received such messages monthly or more often.

Table 1 shows that people who had been exposed to threatening/obscene messages from a partner were more likely: to be young (aged 16–24 (18.4%) or 25–34 (28.7%), p = 0.005), single or divorced (52.7%, p < 0.001), unemployed (5.3%, p < 0.001), and living in rented accommodation (58.0%, p < 0.001) and lower income households (18.7%, p = 0.002). While associated with disadvantage, the experience of threatening/obscene messaging was evident in all groups. The profile of those exposed in the past year was similar to that for those who had ever been exposed, although recent experience was more likely in young. There was no significant association with ethnic group (p = 0.505), although the sample was too small to examine this definitively.

People who received threatening/obscene messages from a current or former partner were more likely to have also experienced all the other types of violence and abuse examined. They were three times more likely than those who had not experienced such messaging to have been abused in childhood emotionally (30.2%, cf. 8.9%), sexually (23.5%, cf. 6.8%), and/or physically (26.5%, cf. 11.4%). They were also about five times more likely to have experienced emotional (59.4%, cf. 8.8%) or physical abuse from a partner (63.1%, cf. 11.4%), or sexual abuse (22.0%, cf. 4.4%) at some point in adulthood. Two-thirds (69.7%) of women and half of men (48.8%) who received threatening/obscene messages had experienced physical partner–violence at some point in adulthood, compared with 14.6% of women and 8.2% of men who had not received threatening/obscene messages.

Rates of CMD were more than twice as high in people who had received threatening/obscene messages (39.2%) than in those who had not (15.2%). This pattern was evident both in women (42.8%, cf. 18.3%) and men (31.5%, cf. 12.0%). Non-suicidal self-harm rates were higher in women (7.1%) and men (3.0%) exposed to such messages, than in women (1.7%) and men (1.3%) not exposed. Suicidal thoughts were three times more common in those exposed to messages (12.6%) than in those who were not (4.1%), while attempted suicide was four times more likely (2.0%, cf. 0.5%). Exposure to abusive messaging was also associated with harmful or dependent use of alcohol. For each of these outcomes, the pattern of association was similar for women and men (see Table 2).

In a model controlling for demographic and socioeconomic factors, people exposed to threatening/obscene messages in the past year had 3.95 times the odds of CMD compared with the rest of the population (95% CI 2.5–6.3, p < 0.001). After further adjustment for exposure to physical violence (in childhood and adulthood), sexual abuse (in childhood and adulthood), and emotional abuse (in childhood only), the association between recent threatening/obscene messages and CMD was attenuated but remained significant (adjusted OR [aOR] 2.20, 1.2–4.0, p = 0.011). Past-year messaging retained an independent association with CMD even when further controlling for experience of any form of emotional abuse from a partner at any point in adulthood (aOR 1.89, 1.0–3.5, p = 0.047), see Table 3 and supplementary materials.

Controlling for demographic and socioeconomic factors, those with past-year exposure to threatening/obscene messages had odds of non-suicidal self-harm in the past year 4.28 times higher than the rest of the population (2.0–9.3, p < 0.001). Further adjustment for physical violence and sexual abuse as a child and adult, as well as for childhood emotional abuse, reduced the odds ratio to 2.95 (1.2–7.1, p = 0.015). In the final model, which additionally adjusted for emotional abuse from a partner across adulthood, the effect size was attenuated (aOR 2.31), although the independent association between past-year threatening/obscene messages and non-suicidal self-harm was not significant at the 95% level (1.0–5.3, p = 0.051).

The odds of having suicidal thoughts in the past year remained 3.76 times higher in those exposed to messages than in those not exposed, when adjusting for demographic and socioeconomic factors (2.1–6.6, p < 0.001). In a model which additionally controlled for all other types of violence recorded in childhood and adulthood, threatening/obscene messages still had a significant independent association with suicidal thoughts (aOR 2.00, 1.1–3.8, p = 0.034).

People with past-year exposure to threatening/obscene messages had odds of attempted suicide in the past year 5.49 times higher than the rest of the population (2.0–15.0, p = 0.001) after adjustment for socioeconomic factors. Further adjustment for physical violence and sexual abuse in childhood and adulthood, as well as for childhood emotional abuse, reduced the odds ratio to 2.94 (0.9–10.1), which did not reach significance at the 95% level (p = 0.086). In the final model, additionally adjusted for any emotional abuse from a partner in adulthood, the effect size remained pronounced (aOR 2.35), although the independent association between messages and suicide attempts remained non-significant at the 95% level (0.8–7.2, p = 0.132).

After adjustment, exposure to abusive messaging in the past year was not associated with harmful or dependent use of alcohol (aOR 1.00, 0.45–2.24, p = 0.996).