Data sources
Data were obtained from the National Board of Health and Welfare and Statistics Sweden. All registers use the ten-digit National Registration Number, assigned to all Swedish residents, which allows linkage studies. Identification numbers were replaced with arbitrary numbers, thereby securing anonymity. The following registers were used: the Swedish Medical Birth Register (MBR), the National Patient Register (NPR), the Swedish Cause-of death register (SCDR), the Multi-Generation Register (MGR), the Register of Total Population (RTP), the Migration Register, the Census of the population and housing and the longitudinal integration database for health insurance and labor market studies (LISA).
The MBR has information on more than 99% of all births in Sweden since 1973. Information is collected from medical records and includes data on maternal diagnoses before and during pregnancy, tobacco and alcohol use, medical drug use, mode of delivery, birth weight, body length, head circumference, duration of pregnancy, Apgar score and infant diagnoses[12].
The NPR has nearly complete nationwide coverage for discharge diagnoses in Sweden based on the International Classification of Diseases (ICD). It has full coverage of all inpatient care in Sweden since 1987. Each record includes admission and discharge dates, the main and secondary discharge diagnoses. Outpatient specialist visits are included since 2001 and the coverage has increased from approximately 75% initially to 87% in 2011[13].
The SCDR includes all individuals who died in Sweden or abroad since 1952 and who were registered in Sweden by the time of death. The data are based on death certificates and provide information on date as well as causes of death using the International Classification of Diseases (ICD)[14].
The RTP includes information about the country of birth.
Data available from the Swedish Multi-Generation Register allow for linkage of individuals born in 1932 and later to parents, siblings, and offspring[15].
The Uppsala Ethics Committee approved the study (Reg. No. 2012/363).
Participants
We identified through MBR and NPR 2454 children born in Sweden between January 1, 1973 and December 31, 2012, who received the diagnosis of UCL, BCL, UCLP or BCLP at birth or prior to 5 years of age.
A comparison group from the population was used: (i) 10 individuals without OFC for each participant with OFC identified (n=24535), matched according to the month and year of birth, sex and county of birth.
Any congenital malformations, deformations and chromosomal abnormalities (ICD 8 and 9 codes 740-759 and ICD-10 codes Q00-Q99, except for the facial cleft codes (749 in ICD-8 and 9 and Q35-Q37 in ICD-10)) were identified as possible syndromic indicators in order to adjust the analyses for non-identified syndromic cases of OFC.
All participants were observed from their date of birth until outcome, emigration, death or end of the study on December 31, 2012.
Exposure
The exposure was a diagnosis of UCL, BCL, UCLP or BCLP as indicated in the MBR or NPR by their International Classification of Diseases ICD-8, ICD-9 or ICD-10 diagnoses. The respective ICD-codes were UCL; 749.10, 749BA and Q36.9A, BCL; 749.11, 749BB and Q36.0A, UCLP; 749.20, 749CC and Q37.5 and BCLP; 749.24, 749CD and Q37.4.
Outcome measures
Information on psychiatric diagnoses and suicide attempts was extracted from the NPR and suicides were extracted from the Swedish Cause of Death Register. We studied the following psychiatric diagnoses (see Table 1 for ICD-codes): any psychiatric disorder (aggregated variable for all the studied disorders in the study); intellectual disability (including mental retardation in ICD-8), speech and language disorders, neurodevelopmental disorders (including autism spectrum disorder (ASD), attention-deficit hyperactivity disorder (ADHD) and behavioural or emotional disorders with onset in childhood), ASD, ADHD, other behavioral/emotional disorders with onset in childhood, a conglomerate of other psychiatric disorders (excluding neurodevelopmental disorders, but including all following), psychotic disorders, bipolar disorder, depression, neurotic, stress related or somatoform disorder, eating disorders, alcohol and substance use disorder, personality disorder, suicide attempt, suicide.
Table 1 ICD-8, 9 and 10 codes for specific psychiatric disorders, as well attempted and completed suicide
|
ICD-8
|
ICD-9
|
ICD-10
|
Any psychiatric disorder
|
290-315
|
290-319
|
F00-F99
|
Intellectual disability
|
310-315
|
317-319
|
F70- F79
|
Speech and language disorders
|
306,00 781,59
|
315D
|
F80 R47
|
Neurodevelopmental disorders
(ASD, ADHD, other behavioral)
|
306,20-308,99
|
299 312-314
|
F84 F90 F91-F98
|
ASD
|
-
|
299
|
F84
|
ADHD
|
-
|
314
|
F90
|
Other behavioral and emotional disorders with onset in childhood
|
306,20-308,99
|
312-313
|
F91-F98
|
Specific learning disorder
|
306,10 781,5 781,6
|
315A 315B 7846
|
F81,0-F81,3 F81,8-F81,9 R48,0
|
Other Psychiatric disorder
(including all below)
|
see all diagnostic codes below
|
see all diagnostic codes below
|
see all diagnostic codes below
|
Psychotic disorders
|
295 297-299 (except 298,00)
|
295 297 298 (except 298A)
|
F20-F29
|
Bipolar disorder
|
296 (except 296,00)
|
296 (except 296B, 296X)
|
F30-F31
|
Depression
|
298,00 300,40 790,20
|
298A 300E 311
|
F32-F33
|
Eating Disorders
|
306,50
|
307B 307F
|
F50
|
Neurotic, stress related or somatoform disorder
|
300 (except 300,40); 305 306,80 306,98 307,99
|
300 (except 300E)
306 307W 308 309
|
F40-F48
|
Alcohol and substance use disorder
|
291 303 294,30 304 971
|
291 303 305A 292 304 305X
|
F10-F16 F18 F19
|
Personality disorder
|
301
|
301
|
F60
|
Suicide attempt
|
E950-E959
|
E950-E959
|
X60-X84
|
Suicide
|
E950-E959
|
E950-E959
|
X60-X84
|
As the ICD-8 and ICD-9 codes for eating disorders were not specific, and we did not want to include early events of feeding difficulties, we only analyzed those receiving a diagnosis of eating disorder from 10 years of age.
Covariates
In order to facilitate the interpretation and comparison of the results of the current study, we used the same strategy in choosing the covariates for the multivariate analyses as in previous research on psychiatric morbidity in OFC[7]. We controlled the multivariate analyses for several confounders: gestational complications and somatic indicators, year and season of birth, sex, congenital malformations or known genetic syndromes, parental psychiatric morbidity and sociodemographic factors.
The following perinatal variables were collected from the MBR; Gestational age at birth, dichotomized into term birth or preterm birth (≥ 37 or <37 gestational weeks). Small for gestational age, defined as less than -2 Standard Deviations. Birth weight was defined as low if <2500 g. Low Apgar score, defined as < 7 at 5 minutes after birth. A binary variable for gestational complications was used in the models (preterm, SGA, low birth weight and low Apgar).
Sociodemographic variables and parental mental health were accessed by linkage of all participants through MGR to the biological parents. Parental age at the time of birth was identified and we used the mean age of the parents or the age of one parent if one was missing for the multivariate analyses.
Maternal country of birth was accessed from the MBR and aggregated across regions: Sweden, other Nordic countries and outside of the Nordic countries. The parental educational level was retrieved from the LISA database and entered into the model as a categorical five level variable according to the Swedish Education Terminology: 9 years, 10-11 years, 12-14 years and >14 years (university). The highest level of education obtained by either of the parents was used in the analysis.
Psychiatric morbidity among parents was defined as at least one psychiatric diagnostic code or a suicide attempt code in the NPR (290-315 in ICD-8, 290-319 in ICD-9 and F00-F98 in ICD-10 or E950-E959 in ICD-8 and ICD-9 and X60-X84 in ICD-10), or a death by suicide in the SCDR. This ordinal variable would take the value of 0, 1 or 2 (number of parents with psychiatric morbidity). We treated the variable as time varying in the analyses.
Statistical analysis
The statistical software Stata v.15 was used for the data analyses[16]. Crude and adjusted Cox proportional hazard regression models were used to investigate hazard ratios and 95% confidence intervals (CI) for the outcomes and age was the underlying time scale. Since the controls cannot be considered as independent from the corresponding case (matched for birth month, sex and county), cluster robust variance-covariance estimation was used[17]. We compared the individuals with BCL with the individuals with UCL and individuals with BCLP with the individuals with UCLP. We also did additional analyses with the different cleft types (UCL, BCL, UCLP and BCLP) and the matched comparison cohort with non-affected individuals.
For the continuous covariates concerning parental age and year of birth, we used restricted cubic spline to avoid forcing the relationship to the outcomes to be linear.
In order to examine a possible moderating effect of sex, we estimated separate Cox models where an interaction term of sex and the exposure was included. The p-value of the interaction term was calculated, except for some of the models that were not possible to estimate, because of too rare outcomes.