This manuscript presents the design and initial outcomes of the New England Family Study's (NEFS) High-Risk Project, one of the few epidemiologically representative cohorts that has prospectively followed a large sample of offspring of parents with both affective and non-affective psychotic disorders from the fetal period forward. The goals of this report are: (1) to describe in some detail the design, data collection methods, and resulting sample of this project; and (2) to prospectively identify and compare rates of childhood neurological impairments among offspring of psychotic and nonpsychotic parents, with a particular emphasis on offspring risk in relation to specific classes of parental psychosis (i.e., affective vs. non-affective psychosis). The investigators identified a pool of 755 parents with potential psychotic disorders, located over 80% of these and confirmed psychotic diagnoses for 212 affected parents and 132 unaffected control parents. At birth, the 259 offspring of parents with psychosis had approximately a twofold increased risk of abnormal neurological functioning compared to offspring of families with no psychotic history. This was most pronounced among the 58 offspring of parents with schizophrenia. Similar trends were observed at ages 1 and 7 years although these did not reach statistical significance. Neither at birth nor at any of the follow-up assessments were the 157 offspring of parents with affective psychosis found to be at elevated risk of neurological impairment. Implications for future research and potential preventive interventions for at-risk individuals are discussed. © 2013 Wiley Periodicals, Inc.
High-risk (HR) studies have proven invaluable over the past 50 years to advance understanding of the biological and social origins and course of schizophrenia and other psychotic disorders [Niemi et al., 2003; Cunningham Owens and Johnstone, 2006]. Alternative designs, such as cohort and conscript studies, are difficult to carry out due to the relatively low incidence of psychotic disorders in the general population, necessitating a large sample of individuals at general risk. An associated challenge is that with such large samples it is difficult and expensive to collect rich and detailed premorbid assessments prior to the onset of later psychotic disorders. In contrast, HR studies—in which a disorder or a condition is studied by selecting individuals at established elevated risk for the disorder [Cornblatt and Obuchowski, 1997]—require smaller sample sizes and allow a wider range of methods to be employed because of reduced expense in sampling. Specifically, the ability to identify individuals at a very early age (sometimes at or even before birth) who are at elevated risk of developing schizophrenia later in life owing to an ill family member, usually parents or siblings, (“family HR” or “genetic HR” design) has supplemented the work from cohort studies, and has permitted focused and detailed assessments of premorbid conditions.
Prior family high-risk (FHR) studies have generally supported the idea of parental mental illness as a substantial risk factor for atypical child development and adjustment. This is especially true for parental psychotic disorders [Hans et al., 2004; Johnstone et al., 2005]. For instance, a number of studies have reported greater rates and greater severity of neurological, motor, and cognitive impairments among offspring of parents with schizophrenia compared to offspring of unaffected parents, including infants [Sameroff et al., 1984; Blennow and McNeil, 1991], younger children [Rieder and Nichols, 1979; Auerbach et al., 1993; McNeil et al., 2003] as well as adolescent and adult offspring [Schubert and McNeil, 2004; Seidman et al., 2006]. Regarding neurological functioning in particular, both the Israeli Kibbutz-City High Risk Study [Marcus et al., 1985] and the Jerusalem Infant Development Study [Hans et al., 1999] reported higher rates among HR offspring. Marcus et al.  found that offspring of parents with schizophrenia had significantly greater deficits in motor coordination, sensory-perceptual signs, and balance. These offspring also exhibited poor right-left orientation and motor overflow. Hans et al.  performed neurobehavioral assessments at ages 17–18 years of 24 offspring of parents with schizophrenia and 16 offspring of unaffected control parents. Among the HR offspring, 42% exhibited impaired neurobehavioral functioning (based on motor and cognitive-attentional variables) compared to 4% of the unaffected offspring.
Less is known about the neurological development of offspring of parents with affective psychosis in comparison to offspring of parents with non-affective psychoses. The Swedish High Risk Study [McNeil et al., 1993] examined a series of neuromotor behaviors and “different facets of general mental development” in relation to the general class of parental psychosis. These authors reported that summary neuromotor deviation scores were significantly higher among offspring of mothers with schizophrenia than their matched control offspring (P = 0.027), but not so for offspring of mothers with affective psychoses and their matched controls. In particular, offspring of mothers with schizophrenia had significantly lower scores on locomotor (P ≤ 0.001) and eye-hand (P ≤ 0.005) measures (reflecting gross and fine motor performance) than their matched controls, differences that were not observed among offspring of mothers with affective psychoses. The results of the study suggested that neuromotor deficits are specific to offspring at HR for schizophrenia as offspring among mothers with affective psychoses did not exhibit similar deviations relative to their controls. In contrast, the Helsinki High-Risk Study examined childhood (age 7–17) neurological soft signs (e.g., “with tics or subthreshold hypotony) in 159 high risk and 99 control offspring [Niemi et al., 2005]. They reported that 10% of the 114 offspring of mothers with nonaffective psychoses and 24% of the 21 offspring of mothers with affective psychosis exhibited neurological soft signs. The rate of neurological soft signs was significantly greater among offspring of mothers with affective psychosis (P = 0.0006)—but not in offspring of mothers with schizophrenia—compared to offspring of healthy control parents. Thus, while prior work tends to suggest more neurological abnormalities in HR offspring of parents with schizophrenia versus affective psychosis, results are not conclusive.
In this manuscript, we present the design and initial outcomes of the New England Family Study's (NEFS) High-Risk Project, one of the few epidemiologically representative cohorts that has prospectively followed a large sample of HR offspring of parents with both affective and non-affective psychotic disorders from the fetal period forward. The goals of this report are: (1) to describe in some detail the design, data collection methods, and resulting sample of the NEFS HR Project; and (2) to prospectively identify and compare rates of childhood neurological impairments among offspring of psychotic and non-psychotic parents, with a particular emphasis on offspring risk in relation to specific classes of parental psychosis (i.e., affective vs. non-affective psychosis).
Participants for this study were selected from the Providence, RI and Boston, MA cohorts of the Collaborative Perinatal Project (CPP) of the National Institute of Neurological and Communicative Disorders and Stroke [Niswander and Gordon, 1972], also known as the NEFS. Pregnant women who were receiving prenatal care at 1 of 12 participating university medical centers were recruited into the CPP study between 1959 and 1966. From the Providence and Boston sites, 17,741 pregnancies (13,464 mothers) were followed prospectively, and events of gestation, labor, delivery, and the neonatal period were systematically assessed. Children's mental, motor, sensory, and physical development were also assessed at 4 and 8 months and 1, 4, and 7 years of age. We refer to the pregnant women and their partners as “Generation 1” or G1 and the study offspring as Generation 2 (G2), following the terminology of Hardy et al. .
Identifying G1 parents with major potential psychotic disorders
We attempted to identify all G1 participants with a history of psychiatric treatment and/or hospitalization (Fig. 1). In a previous study with the Boston cohort, Rieder et al.  had identified 154 families in which either one or both parents had psychiatric diagnoses: 71 had definite psychosis; 46 had possible psychosis; and 37 had other nonpsychotic diagnoses. The 117 families (66 mothers; 51 fathers) with either definite or possible psychosis were included in the potential sample for this study. Subsequent record linkages with psychiatric treatment facilities identified an additional 437 G1 parents who had received psychiatric treatment and/or hospitalization. Of these, 190 parents (159 mothers; 31 fathers) had a linkage diagnosis of either psychosis or bipolar disorder and were included in the potential sample for this study.
Parents with a history of psychiatric treatment and/or hospitalization were identified at various times in the original CPP study (total n = 1,076). At study enrollment, all of the women were asked: (1) “have you ever been a patient in a psychiatric hospital?” and (2) “has the father of this baby ever been a patient in a psychiatric hospital?” Maternal and paternal history of outpatient treatment and treatment for drug or alcohol addiction was also recorded. 164 parents (65 mothers and 99 fathers) had reports of having received inpatient psychiatric treatment; all were included in the potential sample. At the child's 7 year evaluation, study mothers were asked “has anyone in this child's family been treated for a psychiatric condition?” At this assessment, 907 families reported maternal (n = 556), paternal (n = 184), or some combination of family member (n = 167) psychiatric treatment and five mothers were included because they reported having taken anti-psychotic medications at the baseline assessment. Chart notes were reviewed for all of these and a total of 518 G1 parents included in the fielded sample.
In recent follow-up studies with the CPP cohort, there were 22 instances in which a study participant reported that a G1 parent had a history of psychiatric treatment and/or hospitalization or a psychotic diagnosis—these were included in the potential study sample. Finally, based on interviews with a randomly selected sample of potential control subjects, we identified 25 G1 parents who reported psychiatric hospitalization, of which 12 were included in the potential case sample based on review of interview notes.
From these sources, a total of 859 parents were identified with a history of psychiatric treatment and/or hospitalization and whom had some potential indication of a probable or possible psychotic disorder (Fig. 1). We excluded 64 parents whose offspring were not eligible for follow-up in the original CPP (i.e., child was stillborn, aborted, not seen past birth, or adopted away from the project), resulting in 795 eligible G1 participants. Of these, relocation efforts were initiated for 755 participants (Fig. 2).
Located participants were invited to participate in a two-part diagnostic procedure. The first (screening) interview assessed sociodemographic characteristics, medical history, developmental history (e.g., schooling, educational or learning problems, head injuries), and screened for psychiatric symptomatology. Psychiatric symptomatology was assessed using the Quick Diagnostic Interview Schedule [Robins et al., 1989] to screen for Axis I disorders. The Personality Dimensions Questionnaire [Hyler, 1990] was used to screen for Cluster A, Axis II personality disorders (i.e., schizotypal, schizoid, and paranoid personality disorders). Participants were also asked to sign releases for medical records for psychiatric hospitalizations and outpatient treatment. This interview was conducted by systematically trained Bachelor's-level research assistants and then reviewed by four expert doctoral-level diagnosticians to determine whether there was sufficient evidence to indicate potential psychosis and thus to warrant a second interview.
The second interview used the Structured Clinical Interview for Diagnosis (DSM-IV: SCID [First et al., 1996]) and yielded Axis I diagnoses of any form of psychotic, major affective, bipolar or substance use disorders. Family history of psychiatric disorders was evaluated using the Family Interview for Genetic Studies [Maxwell, 1996]. The expert diagnosticians reviewed all of the information collected from both interviews and medical records1, if available, to determine final best estimate diagnoses.
Selection criteria for parents without a history of psychiatric treatment
Once a potential HR parent was identified, matched control parents were selected to be comparable on the following characteristics: number of offspring enrolled in the CPP; patient insurance status (public or private); parent age; ethnicity (Caucasian or other); study site; G2 offspring age, sex, and history of chronic fetal hypoxia. We sought a comparable proportion of comparison offspring with a history of chronic hypoxia to maximize statistical power to examine the independent and joint effects of parental diagnosis and this perinatal risk factor. Offspring were rated positive for a history of chronic perinatal hypoxia if prenatal records indicated one or more of the following: mild to severe preeclampsia; maternal hypertension (diastolic blood pressure during pregnancy of 95 mmHg or greater); maternal hypotension (diastolic blood pressure during pregnancy of less than 60 mmHg); or gestational diabetes (based on insulin therapy, insulin reaction, or blood sugar 200 mg or greater during pregnancy). Eligible controls included all members of the CPP who were not identified as potential psychotic parents and whose records did not indicate a history of psychiatric treatment. A mix of potential cases and controls were released for fieldwork to ensure that interviewers were blind to G1 diagnostic status. If the potential control was interviewed and determined to be ineligible as a control, then a replacement control was released. Interviewers remained blind to potential diagnostic status. All human subject activities were reviewed and approved by multiple Institutional Review Boards at Harvard University and participating hospitals.
Exclusion criteria for the sample of unaffected control G1 parents were the following: history of psychiatric hospitalization; Axis I psychotic disorders; bipolar disorder; recurrent major depression without psychotic features; Axis II Cluster A personality disorders; or genetic disorder with known neurobiological deficits (e.g., Huntington's disease). Unaffected G1 parents were also excluded if there was evidence of psychosis, mania, suicide, or genetic disorder with known neurobiological deficits among their siblings or parents. Figures 1 and 2 document these sampling procedures in detail.
Key measures included in these analyses include parental diagnoses, variables reflecting family demographics (i.e., family socioeconomic status and race) and indices of possible neurological impairment among the G2 offspring. Family socioeconomic status (SES) averaged the percentile scores of parents' education, occupation, and income [Myrianthopoulos and French, 1968]. Maternal race was based on G1 maternal self-report as Caucasian or other. Neurological assessments of G2 offspring were conducted though physician examination at birth, at 1 and 7 years to screen for possible neurological abnormalities though evaluation of the child's central nervous and other body systems. Based on the examination, an overall classification was made of no, possible or certain neurological impairment. For the current analyses, the categories of “possible” and “certain” neurological impairment were combined.
Generalized Estimating Equations [Fitzmaurice et al., 2004] modeling procedures were used to estimate the odds of neurological impairments among G2 offspring relative to G1 parental diagnostic group. All analyses controlled for family SES and maternal age, and accounted for multiple children per family. Rates of neurological impairments among G2 offspring were examined based on assessments at birth, 1 year, and 7 year, as well as a combined model which considered rates of impairments for all time points combined.
Study Sample and Completion Rates
As shown in Figure 2, screening and diagnostic procedures resulted in a sample of 212 parents with DSM-IV psychotic disorders (153 mothers and 59 fathers). This included 84 parents with non-affective psychoses (59 with schizophrenia; 1 with schizophreniform disorder; 5 with delusional disorder; and 19 with psychosis not otherwise specified); 20 with schizoaffective disorder (11 with depressed type; 9 with bipolar type); 96 with affective psychoses including bipolar disorder (n = 48) or major depressive disorder (MDD) with psychosis (n = 48). Based on past literature on familial transmission of schizophrenia and affective psychoses [Faraone and Tsuang, 1985; Kendler et al., 1985; Tsuang, 1993], parents with schizophrenia, schizoaffective disorder of depressed type, delusional disorder, brief psychosis, schizophreniform disorder, and psychosis NOS were classified into one higher order group (schizophrenia psychosis spectrum disorders, subsequently referred to as SPS), and schizoaffective disorder of bipolar type, bipolar disorders with psychosis, and MDD with psychosis were classified into a second group (affective psychoses, subsequently referred to as AP). Twelve parents with a history of psychotic diagnosis which was brief compared to other psychiatric disorders (e.g., major depression or substance disorders) were also included in these two psychotic groups (n = 11 in the affective group; n = 1 in the non-affective group). For 57 subjects, diagnoses were based upon hospital chart information only. In the four instances where both parents had a psychotic condition, the parent with the greater certainty of a psychotic diagnosis was chosen. This resulted in a sample of 116 parents diagnosed with affective and 92 parents with non-affective psychotic disorders.
A sample of 308 parents without a history of psychiatric treatment was identified as potential controls (268 mothers and 40 fathers). Of these, 12 parents were determined to have had a history of psychotic illness and six were not attempted for follow-up. Of the remaining 290 parents, 264 (91%) were definitely located and 45 of these (17%) refused participation. Based on exclusion criteria detailed above, a final sample of 132 parents (119 mothers and 13 fathers) without psychiatric history were included in the unaffected G1 parent group.
Table I shows demographic characteristics of the unaffected and HR samples, including comparisons of participants with different types of psychoses. Significant differences emerged in (a) the number of children per family, where families with a history of schizophrenia or any non-affective psychosis tended to have fewer children than unaffected families, (b) family SES, where families with history of any non-affective psychosis tended to have lower SES than unaffected families, and (c) maternal age at study entry, where mothers with any affective or non-affective psychosis tended to be younger than unaffected mothers.
|G1 diagnostic group||Unaffected||Any psychosis||Any AP||Any NAP||SCZ|
|Parental characteristics||N = 132||N = 208||N = 116||N = 92||N = 57|
|Family SES at study entry; M (SD)||5.7 (2.1)||5.4 (2.0)||5.5 (2.1)||5.2 (1.9)§||5.3 (1.9)|
|Mother's age at study entry; M (SD)||27.1 (6.6)||25.4 (5.6)§||25.4 (5.4)*||25.4 (5.8)*||25.9 (6.4)|
|CPP offspring in family; M (SD)||1.4 (0.8)||1.4 (0.63)||1.4 (0.7)||1.2 (0.5)§||1.2 (0.5)*|
|Offspring characteristics||N = 186||N = 281||N = 167||N = 114||N = 68|
The final sample of 208 families with and 132 families without a psychiatric history had a total of 467 pregnancies enrolled in the CPP (167 offspring of parents with affective psychoses, 114 offspring of parents with non-affective psychoses; and 186 offspring of unaffected parents).
Neurological data were available for 434 offspring at birth; for 408 at 1 year; and for 409 at 7 years of age. As a result, 93% of the sample had at least one, and 77.1% had all three neurological assessments. Families with a history of any psychotic disorder were equally likely as control families to have at least one valid neurological assessment.
Neurological impairment among offspring of parents with psychotic disorder
Table II presents the rates of abnormal neurological assessments relative to parental diagnostic group. When considering all assessments simultaneously (“overall results”), offspring of parents with schizophrenia were observed to have a significantly elevated risk of abnormal neurological functioning compared to the offspring of unaffected parents (odds ratio = 1.98; 95% confidence interval 1.05–3.71). This relative risk was non-significantly elevated for both the broader group of offspring of any non-affective psychoses (OR = 1.63; P = 0.08) and offspring of parents with affective psychoses (OR = 1.44; P = 0.12), and as a consequence among the offspring of all psychoses combined (OR = 1.51; P = 0.06). For both the neonatal and age 7 year assessments, there was a consistent pattern where the rates of neurological impairment were greatest among the offspring of parents with schizophrenia, followed by offspring of any non-affective psychoses, any affective psychoses and lowest among low-risk offspring. These results were most pronounced at birth and when considering all assessments combined. Secondary analyses which considered the temporal patterning of neurological impairments (persistent, late occurring only, early occurring only) revealed no clear relation to parental psychopathology.
|Any affective psychosis||157||—||1.44a,b||0.12||0.90–2.30|
|Any non-affective psychosis||102||—||1.63||0.08||0.95–2.82|
|Any affective psychosis||157||9.6||1.58a||0.29||0.67–3.72|
|Any non-affective psychosis||102||15.7||2.74a||0.02||1.15–6.66|
|Year 1 assessment||408|
|Any affective psychosis||143||13.3||1.84||0.12||0.84–4.03|
|Any non-affective psychosis||94||8.5||.81||0.68||0.30–2.22|
|Year 7 assessment||409|
|Any affective psychosis||143||16.1||1.16||0.63||0.63–2.20|
|Any non-affective psychosis||94||18.1||1.36||0.38||0.68–2.72|
This report provides a detailed summary of the design, assessment procedures and completion rates for the New England Family Study HR Project, initiated by Prof. Ming Tsuang. The utility of the HR design was demonstrated through examination of a simple yet important question: are offspring of parents with psychosis (and certain classes of psychoses) at elevated risk for early neurological impairments? We report that offspring of affected parents were at elevated risk, with such impairments most pronounced at birth. At birth, HR offspring had approximately a twofold increased risk of abnormal neurological functioning compared to offspring of families with no psychotic history. This was most pronounced among offspring of parents with schizophrenia: almost 20% of children from this group exhibited some neurological dysfunction at birth compared to less than 6% of children of unaffected parents. Neither at birth nor at any of the follow-up assessments were offspring of parents with affective psychosis found to be at elevated risk of neurological impairment. While these differences were in the expected direction at the 1 and 7 year assessments, the results did not reach statistical significance. However, when the entire developmental period was examined (birth through 7 years), the results were consistent with the neonatal assessment and elevated odds were observed in relation to parental psychosis (especially schizophrenia).
These results are consistent with prior reports on neurological functioning among offspring of parents with psychoses, in which these problems were most prominent, (1) in children of more severely afflicted parents (i.e., parents with schizophrenia as opposed to any other psychosis, or parents with chronic as opposed to acute schizophrenia) [Rieder and Nichols, 1979; McNeil et al., 1993]; and (2) at birth or in the first months of life, but then gradually diminishing over time [Sameroff et al., 1984; Marcus et al., 1985].
Early characterization of neurological impairments in children of parents with schizophrenia and other psychoses may help identify those who may be at higher than usual risk for subsequent disability. This could help target children who may need to be followed through adolescence and young adulthood allowing for early intervention, if for example, they developed school or neurocognitive problems. Such difficulties may look superficially like a learning disability (LD) or attention deficit disorder (ADD), but the presence of a family history of psychosis combined with neurological or neuropsychological impairments may warrant different interventions than standard.
In later adolescence, during the period when persons “at risk” for psychosis are most likely to develop psychotic illness, the prior presence of a history of neurological dysfunctions and family history of psychosis may heighten clinical attention, particularly if the at-risk person begins to manifest attenuated (mild, subthreshold) positive symptoms [McGlashan and Johannessen, 1996]. These symptoms (“clinical high risk syndrome”) have been shown to predict psychosis, in about 35% of cases within approximately 2.5 years [Cannon et al., 2008]. It is not yet known to what extent the presence of childhood neurological dysfunction may increase risk within this context. We would speculate that this risk is increased as high risk individuals who later convert to psychosis have more severe neuropsychological impairment than non-converters and converters with a positive family history of psychosis have the most severe neuropsychological impairment prior to onset of psychosis [Seidman et al., 2010]. Other publications with this NEFS high-risk project have reported on rates of psychotic disorders in adulthood [Goldstein et al., 2010, 2011]. Consistent with previous high-risk samples, approximately 12% of the high-risk offspring met lifetime diagnostic criteria for a psychotic disorder, a fivefold increase compared to the unaffected control sample. Subsequent analyses will investigate the extent to which early neurological impairments help identify those who will later develop psychotic disorders. With these and related longitudinal investigations, many stimulated by the pioneering work of Dr. Ming Tsuang, we have entered a new phase of early intervention and prevention of psychiatric disorders [Insel, 2006]. As this work progresses, it will be important to integrate knowledge from genetic and clinical high risk approaches as we move in the direction of developing ethically acceptable strategies of early detection and efficacious and non-harmful prevention approaches [Cannon, 2005].
Work for this study was supported by NIMH RO1 MH50647 (1999–2003, Tsuang, P.I.; 2003–2006, Goldstein, P.I.), and the Stanley Medical Research Institute (Buka, P.I.). The authors would like to express their appreciation to the dedicated NEFS research team including Jasmina Burdzovic-Andreas, Ph.D.; Sara Cherkerzian, Sc.D., Lisa Denny, M.D.; Jo-Ann Donatelli, Ph.D., Christiana Provencal, M.A.; Anne Remington, M.A., and to the sustained involvement and many contributions of the NEFS study participants.
Diagnosticians reviewed diagnostic information for 500 participants. Of these reviewed participants, the following types of data were available: 62% interview only; 20% interview and medical records; 18% record linkage diagnosis only.