To the Editor:

I would like to thank Sally McManus and colleagues for this opportunity to engage in a correspondence regarding this important topic of non-suicidal self-injury (NSSI) epidemiology.

I agree with several points shared by McManus and colleagues. Prevalence of several clinical phenomena, such as suicide and depression, have been observed to change at the population level over time (Hedegaard, Curtin, & Warner, Reference Hedegaard, Curtin and Warner2020; Mojtabai, Olfson, & Han, Reference Mojtabai, Olfson and Han2016). McManus and colleagues provide important and unique data that similarly suggest shifts in prevalence of NSSI with time (McManus et al., Reference McManus, Gunnell, Cooper, Bebbington, Howard, Brugha and Appleby2019).

The observation that a relatively older data set was used in my paper is also absolutely correct. The assertion that the year of data collection was not presented in the introduction, however, is not; it is mentioned in the introduction, followed by a citation of McManus, Meltzer, Brugha, and Bebbington (Reference McManus, Meltzer, Brugha and Bebbington2009), and it is mentioned again in the methods section. Beyond indicating the date of data collection, care was taken to specify that the data were from the 2007 Adult Psychiatric Morbidity Survey (APMS), with the aforementioned citation provided, to avoid potential confusion with the 2000 and 2014 surveys of the same name. The 2007 survey data were used because they remain, to my knowledge, the most recent epidemiological data publicly available that would allow for addressing the aims of my paper. The limitations of these data were mentioned in the discussion section, as were the uniquely valuable insight these data nonetheless provide regarding the phenomenology of NSSI. There is also precedence for such use of older epidemiological data. The National Comobidity Survey Adolescent Supplement serves as a prominent example. Data collection was from February 2001 to January 2004. Yet, several important studies on prevalence and/or psychiatric comorbidity have been published after a similarly long period of time, or even longer (e.g. Burstein, Beesdo-Baum, He, & Merikangas, Reference Burstein, Beesdo-Baum, He and Merikangas2014; Nock et al. Reference Nock, Green, Hwang, McLaughlin, Sampson, Zaslavsky and Kessler2013; Paksarian et al. Reference Paksarian, Rudolph, Stapp, Dunster, He, Mennitt and Merikangas2020; Platt, Keyes, McLaughlin, & Kaufman, Reference Platt, Keyes, McLaughlin and Kaufman2019). This reflects the unique value that such epidemiological data continue to provide.

As for being the first epidemiological study of NSSI, were such a claim made of the current study, I agree that it would be indefensible. This claim, however, was never made. Indeed, the second paragraph of the introduction focused entirely on several prior epidemiological studies of this phenomenon. Moreover, in the introduction, I made a qualified statement: my intent was to offer ‘the first epidemiological study to present (i) fine-grained data on sociodemographic and psychiatric correlates of NSSI, (ii) evaluations of NSSI in relation to suicidal ideation and suicide attempts, and (iii) correlates of medical and psychiatric treatment use specifically for NSSI’ (Liu, Reference Liuin press), rather than the first NSSI epidemiological study of any kind. I will briefly elaborate on each of the three qualifiers in turn. First, most, if not all, epidemiological studies have provided some data on sociodemographic correlates. In my study, however, I was interested in presenting unique findings relating to a broad range of sociodemographic characteristics, including sex, race/ethnicity, age, education, marital status, income, poverty, and parental involvement during childhood. Also, to my knowledge, no prior epidemiological study has examined psychiatric disorders in relation to NSSI, outside of a notable recent study with college students (Kiekens et al., Reference Kiekens, Hasking, Claes, Mortier, Auerbach, Boyes and Bruffaerts2018; although here the focus was on the provisional DSM-5 NSSI disorder rather than NSSI occurrence).Footnote ^†Footnote ¹ Second, to my knowledge, no prior epidemiological study has examined NSSI as a predictor of suicidal ideation and behavior. Third, regarding medical and psychiatric treatment for NSSI, McManus et al. (Reference McManus, Gunnell, Cooper, Bebbington, Howard, Brugha and Appleby2019) combined the two in presenting some valuable findings on temporal changes over time and individual characteristics as predictors. In contrast, I sought to observe the important distinction between treatment for physical injury from NSSI (e.g. receiving stitches) and treatment for it as a mental health concern (e.g. receiving psychotherapy for NSSI), and as detailed in my paper, doing so yielded notably different patterns of findings for each.

Finally, I would like to share my view that the primary aims of McManus and colleagues' recent study (McManus et al., Reference McManus, Gunnell, Cooper, Bebbington, Howard, Brugha and Appleby2019) and mine are rather complementary. The former leverages the APMS data sets to provide unique insight on population-level temporal trends over a 15-year period, whereas the latter sought to present an in-depth snapshot characterizing sociodemographic and clinical phenomena associated with NSSI. These two studies therefore collectively provide a more complete understanding of NSSI than would be the case in a single study by itself.