Using Data Linkage to Investigate Inconsistent Reporting of Self-Harm and Questionnaire Non-Response

The objective of this study was to examine agreement between self-reported and medically recorded self-harm, and investigate whether the prevalence of self-harm differs in questionnaire responders vs. non-responders. A total of 4,810 participants from the Avon Longitudinal Study of Parents and Children (ALSPAC) completed a self-harm questionnaire at age 16 years. Data from consenting participants were linked to medical records (number available for analyses ranges from 205–3,027). The prevalence of self-harm leading to hospital admission was somewhat higher in questionnaire non-responders than responders (2.0 vs. 1.2%). Hospital attendance with self-harm was under-reported on the questionnaire. One third reported self-harm inconsistently over time; inconsistent reporters were less likely to have depression and fewer had self-harmed with suicidal intent. Self-harm prevalence estimates derived from self-report may be underestimated; more accurate figures may come from combining data from multiple sources.


INTRODUCTION
Community studies of self-harm are vital as the majority of self-harm episodes do not present to clinical services (Hawton, Rodham, Evans, & Weatherall, 2002;Kidger, Heron, Lewis, Evans, & Gunnell, 2012;Ystgaard et al., 2009). However, such studies are subject to a number of limitations such as misreporting and nonresponse (Grimes & Schulz, 2002), which can lead to bias in estimates of prevalence and measures of association. The extent to which this occurs in the case of self-harm is not currently known.
Non-response and loss to follow-up occur more frequently among individuals with particular characteristics (Kidger et al., 2012;Wolke et al., 2009). For example, in the Early Developmental Stages of Psychopathology Study (Christl, Wittchen, Pfister, Lieb, & Bronisch, 2006), participation rates at follow-up were lower among those who had attempted suicide compared to those without suicidal thoughts or attempts. This skewed pattern of participation would have led to underestimates of the prevalence of suicide attempts.
In addition, as information on selfharm is typically collected retrospectively via self-report, the accuracy of responses may be affected by issues such as denial, reinterpretation, problems with recall, current mood, or by misinterpretation of the study questions (Velting, Rathus, & Asnis, 1998). There is also evidence to suggest that concerns over social desirability may encourage under-reporting, as adolescents have been found to report suicide attempts two to three times more frequently under conditions of anonymity (Safer, 1997).
Whereas previous studies have typically compared self-report questionnaire and interview responses, the present study compares self-reported self-harm with data from medical records. This data is external and objective, although it cannot be considered to be free from error. We linked data from medical records with data reported by participants in the Avon Longitudinal Study of Parents and Children, a longitudinal population-based birth cohort (Boyd et al., 2013). Our aims were to: 1. Investigate whether the prevalence of self-harm recorded in medical records differs between responders and non-responders to the self-harm questionnaire. 2. Investigate the level of agreement between self-report and medically recorded self-harm events. 3. Examine consistency in the reporting of self-harm in ALSPAC over time, by comparing questionnaire responses at age 16 and 18 years. 4. Identify characteristics associated with inconsistent reporting of self-harm over time.
We hypothesize that the prevalence of self-harm recorded in medical records will be higher among questionnaire nonresponders than responders, and that recall of self-harm episodes over time will be most consistent in individuals with more severe mental health problems=selfharm.  , 1991and December 31, 1992(Boyd et al., 2013. Of the 14,062 live births, 13,798 were singletons= first-born of twins and were alive at 1 year of age. Participants have been followed up since recruitment through regular questionnaires and research clinics. Detailed information about ALSPAC is available on the study website (http://www.bristol.ac.uk/ alspac), which includes a fully searchable data-dictionary of available data (http:// www.bris.ac.uk/alspac/researchers/dataaccess/data-dictionary). Self-harm was assessed via self-report questionnaire at age 16 years (mean age of respondents 16 years 8 months, standard deviation [SD] approximately 3 months). The postal questionnaire was sent to 9,383 participants of whom 4,855 (51.7%) returned it and 4,810 completed the self-harm items (Kidger et al., 2012). Ethical approval for the study was obtained from the ALSPAC Law and Ethics committee and local research ethics committees (NHS Haydock REC: 10= H1010=70).
Linkage. The Health and Social Care Information Centre (HSCIC) linked ALSPAC participants with the NHS Central Register, with a 99% match rate (Boyd et al., 2013); this was done on the basis of NHS ID number, name, date of birth, and postcode using deterministic linkage.
When the ALSPAC children reached adulthood (age 18), they were invited to enroll in the study in their own right and to consent to the extraction and use of their health records. Through the Project to Enhance ALSPAC through Record Linkage (PEARL) http://www.bristol. ac.uk/alspac/participants/playingyourpart/ information and consent forms were posted to 12,385 of the participants eligible to be included in this investigation (singletons=first born twins from the ALSPAC core enrolled sample who were alive at 1 year. See Figure 1). Of those invited to consent (n ¼ 12,385), 3,027 (24.4%) consented to data linkage by the study cut-off date, 8,905 (71.9%) did not respond to the consent request, and 82 (0.7%) returned an incomplete consent form. Only 371 (3.0%) declined to consent.
The Hospital Episode Statistics Database (HES). The HES database (Copyright # 2012, re-used with the permission of The Health and Social Care Information Centre. All rights reserved) contains information about hospital presentations and admissions for all NHS hospitals in England; it contains admissions data from 1989 onwards, outpatient data from 2003 onwards and A&E data from 2007 onwards (http://www.hscic.gov.uk/hes). Of the 3,027 individuals who consented to data linkage (see above) 2,957 individuals (97.7%) had an existing linkage to the NHS central register, which in turn provided a means to identify the individuals' secondary care records contained in the HES database. The remaining 70 cases were linked to HES using NHS ID number, name, and date of birth. In this scenario ''linkage'' refers to the process of testing if the ALSPAC participants had any HES records rather than the actual identification and extraction of a record. We make this distinction as some individuals will genuinely not have any HES records, while others may have a HES record which we failed to identify during the linkage process. In March 2013 the NHS Health and Social Care Information Centre (HSCIC) extracted the hospital admissions records of 2,988 participants, although we consider the denominator to be the 3,027 cases tested for linkage.
The Clinical Practice Research Datalink (CPRD). The CPRD is an anonymized database of primary care records of around 5 million ($8%) patients in the UK. Linkage between ALSPAC and the CPRD was conducted by the NHS Information Centre (NHS IC) as a trusted third party. With approval from the NIGB Ethics and Confidentiality Committee, the NHS IC identified ALSPAC eligible individuals who also appeared in the CPRD, and sent an anonymized linking dataset to be stored securely at the CPRD where the data were merged and analyzed. This particular linkage does Consistency in Reports of Self-Harm in Adolescence not require consent above and beyond the consent obtained for participation in ALSPAC. However, any participants who did not agree to their health records being extracted (via the PEARL consent request described above) were excluded (n ¼ 3).
Of the live births linked by the NHS IC that appeared in the CPRD, 520 were in the sub-sample eligible for this investigation (singletons=first born twins from the ALSPAC core enrolled sample who were alive at 1 year). The sample was further restricted to individuals who were registered with a CPRD-contributing practice for the entire period between age 10 and 17 years (n ¼ 205) (Figure 2), to ensure that there were no breaks in the patients' records. We did not examine CPRD records before the age of 10 years, as self-harm before this age is rare.

Measures
History of self-harm was assessed in the ALSPAC cohort, the Hospital Episode Statistics Database (Secondary Care) and the Clinical Practice Research Datalink (Primary Care). The methods of assessment for each data source are described below. Data on psychosocial characteristics were also collected in ALSPAC.
The Avon Longitudinal Study of Parents and Children (ALSPAC). The self-harm questions used in the age 16 self-report questionnaire were based on those used in the CASE

B. Mars et al.
study (Madge et al., 2008). Participants who responded positively to the item ''have you ever hurt yourself on purpose in any way (e.g., by taking an overdose of pills or by cutting yourself)?'' were classified as having a lifetime history of self-harm. Those who answered ''yes'' to having self-harmed were then asked further closed response questions, including how long ago they last hurt themselves (in the last week, more than a week ago but in the last year, more than a year ago), the reasons for self-harm the last time they hurt themselves on purpose (six response categories), and whether they had ever seriously wanted to kill themselves when self-harming (Kidger et al., 2012). Participants were classified as having a lifetime history of suicidal selfharm if they selected ''I wanted to die'' as a reason for harming themselves on the most recent occasion, or if they reported they had ever seriously wanted to kill themselves when self-harming (Mars et al., 2014).
The same question was used to assess lifetime self-harm at age 18 years, using the self-administered computerized version of the Clinical Interview Schedule-Revised (CIS-R) (Lewis, Pelosi, Araya, & Dunn, 1992). There is close agreement between the self-administered computerized version and the interviewer administrated versions of the CIS-R (Bell, Watson, Sharp, Lyons, & Lewis, 2005;Lewis, 1994;Patton et al., 1999).
Psychosocial Characteristics. We examined key psychosocial characteristics assessed previously in ALSPAC to identify factors associated with inconsistencies in reporting self-harm over time. The following variables were used: (1) participant's gender, (2) ethnicity, (3) parent social class (professional=managerial or other occupations; the highest of maternal or paternal social class was used), (4) highest maternal educational attainment (less than O-level, O-level, A-level, or university degree) measured during pregnancy (O-levels and A-levels are school qualifications taken around age 16 and 18 years respectively), (5) child IQ assessed using the Wechsler intelligence test for children (WISC-III) (Wechsler, 1991) at age 8 years (6) depression symptoms, assessed at age 16 and 18 years using the short Moods and Feelings Questionnaire (SMFQ), a score of 11 or more on the SMFQ was taken as indicative of depressive symptoms (Patton et al., 2008) and (7) depressive disorder, assessed at age 18 years using the CIS-R.
The Hospital Episode Statistics Database (HES). We used an extract of the HES data including hospital admissions for self-harm (ICD 10 codes Y10-Y34, X60-X84 and X40-X49), A&E attendances for self-harm (A&E diagnostic codes 141=142 ''poisoning (inc overdose) due to prescriptive=proprietary drugs,'' or reason for A&E attendance coded as ''deliberate self-harm'') and hospital admissions for a mental health condition(s) (ICD-10 codes F00-F99). Further details can be found in Appendix 1. While X40-X49 are coded as accidental poisoning, previous studies indicate that they are also used for self-harm. The date of hospital attendance was crossreferenced with the date of questionnaire completion to identify whether events occurred before or after completion of the self-harm questionnaire. Although A&E data is recorded in HES, it is only available from 2007 onwards and is likely to be under-reported. For example, in the extracted data, all but two self-harm hospital admissions were recorded as having come via A&E (the remaining two admissions were emergency referrals by GP); however, two-thirds of hospital admissions had no corresponding A&E record for self-harm. For this reason, we have focused primarily on hospital admissions data in this paper, as this is known to be more complete. The findings for A&E Consistency in Reports of Self-Harm in Adolescence only data are also presented, but need to be interpreted with caution.
The Clinical Practice Research Datalink (CPRD). Cases of self-harm occurring in the CPRD until December 31, 2011 were identified using appropriate Read codes for attempted suicide and self-harm (see Appendix 2) (Thomas et al., 2013).

Analysis Plan
Non-Response. We examined whether there was an association between questionnaire response and medically recorded self-harm by comparing the prevalence of self-harm in HES and the CPRD among those who completed and did not complete the selfharm questionnaire at age 16 years. Agreement Between Self-Report and Medically Recorded Self-Harm Events. We compared self-reported self-harm episodes with events recorded in HES and the CPRD, in order to identify instances in which self-harm was inconsistently reported. Consistency in Self-Report Over Time. We investigated inconsistency in reporting of lifetime self-harm over time between age 16 and 18 years in ALSPAC cohort. Participants who reported no self-harm, or reported self-harm for the first time at age 18 years were excluded from these analyses.
Characteristics associated with inconsistent reporting of self-harm over time were also examined using logistic regression.

Self-Harm
HES. Of the 3,027 ALSPAC participants tested for linkage with HES (hospital records), 54 (1.8%) had one or more self-harm events recorded in HES, including 41 participants with at least one recorded hospital admission for self-harm, and 18 (0.6%) with at least one recorded ''A&E only'' attendance for self-harm (i.e., A&E attendance without subsequent hospital admission). It is notable that 66% of individuals who were admitted to hospital following self-harm had no corresponding A&E record for self-harm. Eighty-two (2.7%) had at least one hospital admission for a mental health condition recorded in HES. Of the 3,027 individuals tested for linkage, 2,363 (78.1%) completed the self-harm questionnaire at age 16 years.
CPRD. Of the 205 ALSPAC participants registered with a CPRD contributing practice between age 10 and 17 years, 64 (31.2%) completed the self-harm questionnaire at age 16 years. Only 6 participants (2.9%) had a relevant self-harm Read code recorded in the CPRD.

Non-Response
HES. The prevalence of hospital admissions for self-harm and mental health conditions recorded in HES was higher among those who did not complete the self-harm questionnaire at age 16 years than among those who did ( CPRD. Two of the 6 individuals with a self-harm Read code recorded in the CPRD completed the age 16 self-harm questionnaire. There was no evidence of a difference in prevalence between questionnaire responders and non-responders (2.8% in non-responders vs. 3.1% in responders, B. Mars et al. difference ¼ 0.3%, 95% CI À4.8-5.4%, P ¼ 0.910). These findings need to be interpreted with caution, given the small number of ALSPAC individuals with a self-harm Read code recorded in the CPRD (n ¼ 6).

Agreement Between Self-Report and Medical Records
HES. Of the 2,363 individuals tested for linkage who completed the self-harm Consistency in Reports of Self-Harm in Adolescence questionnaire at age 16 years, 419 (17.7%) reported a history of self-harm. Only 12 (2.9%; 95% CI 1.5-5.9%) of these episodes were recorded in HES. There were 15 self-harm hospital attendances recorded in HES prior to completion of the self-harm questionnaire (12 admissions and 3 A&E only attendances). Three (20%; 95% CI 4-48%) of these episodes were not reported by ALSPAC participants on the questionnaire (1=12 admissions and 2=3 A&E only attendances).

CPRD.
Both of the self-harm events recorded in the CPRD were reported by participants on the self-harm questionnaire; however, neither participant reported having sought help for self-harm from their GP (a consultation with the GP would be necessary in order for a self-harm Read code to be recorded in the CPRD).

Consistency of Reporting of Self-Harm Over Time
Five hundred and eighty nine individuals reported lifetime self-harm at age 16 years and provided information on self-harm at age 18 years. Of these, 385 (65.4%) reported self-harm consistently at both time points, and 204 individuals (34.6%) reported self-harm inconsistently, i.e., reported lifetime self-harm at age 16 years but not at age 18 years.
Characteristics Associated With Consistency in Reporting of Self-Harm Over Time. Compared with those who reported self-harm consistently over time, those who reported self-harm inconsistently were less likely to have evidence of depression at age 16 and 18 years, were less likely to have self-harmed in the year prior to the age 16 year questionnaire, and were less likely to have harmed with suicidal intent by age 16 years (Table 3). There was little evidence for differences according to gender, social class, IQ, maternal education or ethnicity (Table 3).

Main Findings
This study is, as far as we are aware, the first to examine whether the prevalence of medically recorded self-harm differs from prevalence determined by questionnaire response in a community-based sample of adolescents. We also investigated the level of agreement between self-reported self-harm history and data obtained from medical records.
We found some evidence for both selective non-participation of individuals with self-harm, and for discrepancies between self-reported and medically recorded self-harm episodes; approximately one-fifth of self-harm events recorded in HES (hospital admissions or A&E presentations) were not reported by participants on the questionnaire. Taken together, these findings suggest that prevalence estimates derived from self-report may underestimate the true rate of adolescent self-harm in the community.
We additionally examined the consistency of self-reported self-harm over time and found that over a third of respondents who reported self-harm at age 16 years said they had never self-harmed when asked at age 18 years. Those who reported selfharm inconsistently over time were less likely to have to have depressive disorder, less likely to have harmed in the year prior to the age 16 year questionnaire and were less likely to have self-harmed with suicidal intent.

Strengths and Limitations
ALSPAC is a large, population-based study, which is important, given that less than 20% of adolescents who self-harm B. Mars et al. present to medical services (Hawton et al., 2002;Kidger et al., 2012). We investigated the level of agreement in reports of self-harm both across different sources (self-report and medical records) and over time.
The findings need to be interpreted in light of several limitations. First, we were only able to compare reports among those who had been admitted to hospital or had consulted with their GP. We were also only able to examine self-harm hospital admissions among those who had consented to data linkage (24% of the sub-sample invited to consent) and GP events for those in the CPRD between age 10 and 17 years (1.5% of the sub-sample of 13,798 included in this investigation). These sub samples with available linked records may not be representative of the whole ALSPAC cohort. The issue of required consent has the potential to induce bias in our findings, however using questionnaire data we found little evidence of an association between self-harm and consent to data linkage and so this is unlikely to be a problem in this study. Second, it is likely that cultural differences influence self-reporting of self-harm. The degree of stigma associated with mental illness and self-harm varies around the world (Abdullah & Brown, 2011;Evans-Lacko, Brohan, Mojtabai, & Thornicroft, 2012;Reynders, Kerkhof, Molenberghs, & Van Audenhove, 2014), therefore findings from our study may not be generalizable outside a UK context. Third, the number of individuals with self-harm recorded in their medical records was small, particularly in the CPRD. This precluded our ability to examine characteristics associated with inconsistent reporting, and limited power to detect differences between questionnaire responders and non-responders. Findings therefore need to be interpreted with caution, and require replication in a larger sample. It is possible that some episodes of self-harm may not have been recorded in the CPRD (Thomas et al., 2013), or may have been missed (i.e., if documented as a free text response rather than a Read code). Fourth, when extracting data from the HES database, we included codes related to accidental poisoning (ICD 10 codes X40-X49) as these codes are often used to indicate self-harm. While some may be true instances of accidental selfpoisoning, this is unusual in adolescence.
Finally, self-harm in ALSPAC was assessed via self-report questionnaire at age 16 years and via a self-administered computerized assessment at age 18 years. Although the question used at both time points was identical, the difference in setting may have contributed to the discrepancies in reporting found in this study.

Comparison With Previous Research
Previous studies investigating inconsistency in reporting of self-harm have typically relied on comparisons between interview and questionnaire responses (Bjärehed et al., 2013;Ougrin & Boege, 2013;Ross & Heath, 2002;Velting et al., 1998). Lower rates of self-harm are usually found when using interview as opposed to questionnaire measures (Evans, Hawton, Rodham, Psychol, & Deeks, 2005). However, the absence of a gold standard assessment for self-harm means that it is not possible to identify which of these measurement approaches is more accurate-the ability to ask additional clarification questions could help to eliminate false positives that arise from inaccurate self-reports (Hawton et al., 2002;Ross & Heath, 2002;Velting et al., 1998), but it is also possible that the loss of anonymity found with interview assessments may result in under-reporting of self-harm (Safer, 1997).
In the Early Developmental stages of Psychopathology Study, Christl et al. (2006) found some evidence for selective Consistency in Reports of Self-Harm in Adolescence non-response as those who reported suicide attempts at baseline were at least 1.6 times more likely to drop out of the study than those without suicidal thoughts or behavior. The use of data linkage allows us to extend this work by objectively comparing the prevalence of self-harm among questionnaire responders and nonresponders. There was also some evidence for inconsistency between self-reported and medically recorded self-harm. Possible reasons for discrepancies include concerns over stigma, denial, or problems with recall. Individuals may also suppress painful memories such as self-harm or suicidal ideation, which has been suggested as a possible adaptive defensive mechanism (Goldney, Winefield, Winefield, & Saebel, 2009;Klimes-Dougan, Safer, Ronsaville, Tinsley, & Harris, 2007).
Our finding that a third of adolescents were discrepant in their reporting of lifetime self-harm over time is lower than the proportion found by Hart et al. (2013) (approximately two thirds disrepant 1 year after reporting a self-harm event) but similar to findings of other previous longitudinal research (Eikelenboom et al., 2014, Christl et al., 2006, all of which investigated reporting of suicide attempts. Inconsistent reporting has also been shown for other stigmatized behaviors such as drug use (Percy, McAlister, Higgins, McCrystal, & Thornton, 2005). We extend this research by examining consistency in reporting of self-harm regardless of suicidal intent, and by examining various characteristics associated with discrepant reporting. Similar to Christl et al. (2006) and Eikelenboom et al. (2014), we found greater consistency in reporting among those with psychopathology. We also found individuals were more likely to report self-harm consistently if they had harmed with suicidal intent during their lifetime, and if they had self-harmed in the year prior to questionnaire completion. This could suggest that more severe self-harm episodes and those that are more recent are more likely to be recalled by participants and may be less subject to reinterpretation. However, in their investigation of suicide attempts in adults, Eikelenboom et al. (2014) found no association between consistency in reporting and the recency of self-harm at baseline. It is also possible that individuals with psychopathology and those who have harmed with suicidal intent may be more likely to continue to self-harm as adults. The reasons for discrepant reporting require further investigation and could include denial, errors in recall, or reinterpretation of the self-harm event. Reports may also be influenced by current mood state, for example depressed mood could lead to enhanced recall of negative events, such as self-harm. Unfortunately, it is not possible to determine in this, or other studies, which of the assessments is more accurate (i.e., whether the first reporting of self-harm is a false positive or whether the second reporting is a false negative).
It is also important to note that while selective non-participation of those with self-harm and inconsistent reporting could result in distorted prevalence estimates, this does not necessarily lead to biased estimates of associations between self-harm and exposure variables (Wolke et al., 2009). Further research is planned to investigate this issue in more detail within the ALSPAC cohort.

CONCLUSION
In our analyses of the ALSPAC cohort, we have shown that self-harm prevalence estimates derived from self-report are affected by non-response and inconsistent reporting, and likely underestimate the true level of adolescent self-harm in the community. Our findings require replication, but suggest benefits of combining self-report self-harm data with data from medical records. To maximize the potential for B. Mars et al. this approach would require complete coverage of medical records for the sample in question. In practice achieving this may be restricted by governance requirements based on concerns around the protection of privacy with regard to sensitive information in the situation where individuals have not provided explicit consent. Such concerns may be offset by evidence that data-linkage as we describe here can improve the validity of medical research and thus enhance the potential of research to improve the public good.
David Gunnell, School of Social and Community Medicine, University of Bristol, Bristol, UK. Correspondence concerning this article should be addressed to Dr. Becky Mars,School of Social and Community Medicine,University of Bristol,Oakfield House,Bristol BS8 2BN, We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses.
This publication is the work of the authors who serve as guarantors for the contents of this paper. The study sponsor had no further role in the study design and collection, analysis, and interpretation of data or in the writing of the article and the decision to submit it for publication.  Consistency in Reports of Self-Harm in Adolescence APPENDIX 1 List of ICD-10 Codes Used to Identify Hospital Admissions for Non-fatal Self-harm in the Hospital Episodes Statistics database (HES) ICD-10 code Description

X40
Accidental poisoning by and exposure to nonopioid analgesics, antipyretics and antirheumatics X41 Accidental poisoning by and exposure to antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs, not elsewhere classified X42 Accidental poisoning by and exposure to narcotics and psychodysleptics [hallucinogens], not elsewhere classified X43 Accidental poisoning by and exposure to other drugs acting on the autonomic nervous system X44 Accidental poisoning by and exposure to other and unspecified drugs, medicaments and biological substances X45 Accidental poisoning by and exposure to alcohol X46 Accidental poisoning by and exposure to organic solvents and halogenated hydrocarbons and their vapors X47 Accidental poisoning by and exposure to other gases and vapors X48 Accidental poisoning by and exposure to pesticides X49 Accidental poisoning by and exposure to other and unspecified chemicals and noxious substances X60 Intentional self-poisoning by and exposure to nonopioid analgesics, antipyretics and antirheumatics X61 Intentional self-poisoning by and exposure to antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs, not elsewhere classified X62 Intentional self-poisoning by and exposure to narcotics and psychodysleptics [hallucinogens], not elsewhere classified X63 Intentional self-poisoning by and exposure to other drugs acting on the autonomic nervous system X64 Intentional self-poisoning by and exposure to other and unspecified drugs, medicaments and biological substances X65 Intentional self-poisoning by and exposure to alcohol X66 Intentional self-poisoning by and exposure to organic solvents and halogenated hydrocarbons and their vapors X67 Intentional self-poisoning by and exposure to other gases and vapors X68 Intentional self-poisoning by and exposure to pesticides X69 Intentional self-poisoning by and exposure to other and unspecified chemicals and noxious substances X70 Intentional self-harm by hanging, strangulation and suffocation X71 Intentional self-harm by drowning and submersion X72 Intentional self-harm by handgun discharge Suicide þ self-inflicted poisoning by solid=liquid substances TK00.00 Suicide þ self-inflicted poisoning by analgesic=antipyretic TK01.00 Suicide þ self-inflicted poisoning by barbiturates TK01000 Suicide and self-inflicted injury by amylobarbitone TK01100 Suicide and self-inflicted injury by barbitone TK01400 Suicide and self-inflicted injury by phenobarbitone TK02.00 Suicide þ self-inflicted poisoning by other sedatives=hypnotics TK03.00 Suicide þ self-inflicted poisoning tranquillizer=psychotropic TK04.00 Suicide þ self-inflicted poisoning by other drugs=medicines TK05.00 Suicide þ self-inflicted poisoning by drug or medicine not otherwise specified TK06.00 Suicide þ self-inflicted poisoning by agricultural chemical TK07.00 Suicide þ self-inflicted poisoning by corrosive=caustic substance TK0z.00 Suicide þ self-inflicted poisoning by solid=liquid substance not otherwise specified TK1.00 Suicide þ self-inflicted poisoning by gases in domestic use TK10.00 Suicide þ self-inflicted poisoning by gas via pipeline TK11.00 Suicide þ self-inflicted poisoning by liquefied petrol gas (Continued ) B. Mars et al.

APPENDIX 2. Continued
Read code Description TK1y.00 Suicide and self-inflicted poisoning by other utility gas TK1z.00 Suicide þ self-inflicted poisoning by domestic gases not otherwise specified TK2.00 Suicide þ self-inflicted poisoning by other gases and vapors TK20.00 Suicide þ self-inflicted poisoning by motor vehicle exhaust gas TK21.00 Suicide and self-inflicted poisoning by other carbon monoxide TK2z.00 Suicide þ self-inflicted poisoning by gases and vapors not otherwise specified TK3.00 Suicide þ self-inflicted injury by hang=strangulate=suffocate TK30.00 Suicide and self-inflicted injury by hanging TK31.00 Suicide þ self-inflicted injury by suffocation by plastic bag TK3y.00 Suicide þ self-inflicted injury by other means than hang=strangle=suffocate TK3z.00 Suicide þ self-inflicted injury by hang=strangle=suffocate not otherwise specified TK4.00 Suicide and self-inflicted injury by drowning TK5.00 Suicide and self-inflicted injury by firearms and explosives TK51.00 Suicide and self-inflicted injury by shotgun TK52.00 Suicide and self-inflicted injury by hunting rifle TK54.00 Suicide and self-inflicted injury by other firearm TK5z.00 Suicide and self-inflicted injury by firearms=explosives not otherwise specified TK6.00 Suicide and self-inflicted injury by cutting and stabbing TK60.00 Suicide and self-inflicted injury by cutting TK60100 Self-inflicted lacerations to wrist TK60111 Slashed wrists self-inflicted TK61.00 Suicide and self-inflicted injury by stabbing TK6z.00 Suicide and self-inflicted injury by cutting and stabbing not otherwise specified TK7.00 Suicide and self-inflicted injury by jumping from high place TK70.00 Suicide þ self-inflicted injury-jump from residential premises TK71.00 Suicide þ self-inflicted injury-jump from other manmade structure TK72.00 Suicide þ self-inflicted injury-jump from natural sites TK7z.00 Suicide þ self-inflicted injury-jump from high place not otherwise specified TKx.00 Suicide and self-inflicted injury by other means TKx0.00 Suicide þ self-inflicted injury-jump=lie before moving object TKx0000 Suicide þ self-inflicted injury-jumping before moving object TKx1.00 Suicide and self-inflicted injury by burns or fire TKx2.00 Suicide and self-inflicted injury by scald TKx3.00 Suicide and self-inflicted injury by extremes of cold TKx4.00 Suicide and self-inflicted injury by electrocution TKx5.00 Suicide and self-inflicted injury by crashing motor vehicle TKx6.00 Suicide and self-inflicted injury by crashing of aircraft TKx7.00 Suicide and self-inflicted injury caustic substance, excluding poison TKxy.00 Suicide and self-inflicted injury by other specified means TKxz.00 Suicide and self-inflicted injury by other means not otherwise specified (Continued ) Consistency in Reports of Self-Harm in Adolescence