- Email: [email protected]
Analysis of risk factors for schizophrenia with two different case definitions: A nationwide register-based external validation study
Schizophrenia Research 162 (2015) 74–78
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Analysis of risk factors for schizophrenia with two different case definitions: A nationwide register-based external validation study☆ Holger J. Sørensen a,e,⁎, Janne T. Larsen b,e, Ole Mors d,e, Merete Nordentoft a,e, Preben B. Mortensen b,c,e, Liselotte Petersen b,e a
Psychiatric Centre Copenhagen, Capital Region of Denmark, University Hospital of Copenhagen, Bispebjerg Bakke 23, 2400 NV, Denmark National Centre for Register-based Research, Aarhus University, Business and Social Sciences, Aarhus, Denmark Centre for Integrated Register-based Research, CIRRAU, Aarhus University, Aarhus, Denmark d Research Department P, Aarhus University Hospital, Risskov, Denmark e The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark b c
a r t i c l e
i n f o
Article history: Received 9 October 2014 Received in revised form 29 December 2014 Accepted 8 January 2015 Available online 23 January 2015 Keywords: Schizophrenia Case definition External validation Register-based studies
a b s t r a c t Different case definitions of schizophrenia have been used in register based research. However, no previous study has externally validated two different case definitions of schizophrenia against a wide range of risk factors for schizophrenia. We investigated hazard ratios (HRs) for a wide range of risk factors for ICD-10 DCR schizophrenia using a nationwide Danish sample of 2,772,144 residents born in 1955–1997. We compared one contact only (OCO) (the case definition of schizophrenia used in Danish register based studies) with two or more contacts (TMC) (a case definition of at least 2 inpatient contacts with schizophrenia). During the follow-up, the OCO definition included 15,074 and the TMC 7562 cases; i.e. half as many. The TMC case definition appeared to select for a worse illness course. A wide range of risk factors were uniformly associated with both case definitions and only slightly higher risk estimates were found for the TMC definition. Choosing at least 2 inpatient contacts with schizophrenia (TMC) instead of the currently used case definition would result in almost similar risk estimates for many well-established risk factors. However, this would also introduce selection and include considerably fewer cases and reduce power of e.g. genetic studies based on register-diagnosed cases only. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The decision on when to define a case of schizophrenia in registerbased research is guided by several considerations. In Denmark, one contact to the secondary health care system with schizophrenia has been viewed as sufficient to define a case in register-based research. To increase the diagnostic validity, by minimizing false-positive diagnoses, researchers using the Swedish hospital discharge register sometimes employ a definition requiring a diagnosis of schizophrenia on two separate treatment inpatient contacts (Lichtenstein et al., 2006). However, a case definition of schizophrenia demanding two or more contacts (TMC) could lead to increased risk of selection. On the other hand, the advantage of the TMC definition is reduced heterogeneity among cases and a more homogeneous sample suited for e.g. biological studies (Fazel et al., 2009; Ripke et al., 2013; Ruderfer et al., 2014;
☆ Presented at The 4th Biennial Schizophrenia International Research Society Conference, 5–9 April 2014 in Florence, Italy. ⁎ Corresponding author at: Mental Health Centre Copenhagen, Bispebjerg Bakke 23, 13A, DK 2400 NV, Copenhagen, Denmark. E-mail address: [email protected] (H.J. Sørensen).
http://dx.doi.org/10.1016/j.schres.2015.01.018 0920-9964/© 2015 Elsevier B.V. All rights reserved.
Purcell et al., 2014; Rees et al., 2014; Szatkiewicz et al., 2014). Conversely, the potential disadvantage of the OCO (one case only) definition used in Denmark is a higher risk of false-positive diagnoses. There has been several reports (Dalman et al., 2002; Ekholm et al., 2005; Uggerby et al., 2013) suggesting that the validity of a schizophrenia diagnosis is good in Swedish as well as in Danish nationwide registers. However, since there are no supportive diagnostic tests for schizophrenia, it may be important, when designing e.g. future genetic studies based on register-based samples only, to consider when to use alternative ascertainment or not. In large register-based studies, sample size and effects of common and well-established risk factors could potentially depend on the case definition of schizophrenia. To study this further, we here examine the TMC definition versus the OCO definition in an external validation study. We decided to use well-established risk factors for schizophrenia as external validators as suggested previously (Robins and Guze, 1970). These included family history of mental illness (Mortensen et al., 1999), urban place of birth (Pedersen and Mortensen, 2001), unknown parent status (Mortensen et al., 1999), advanced paternal age (Sorensen et al., 2014; McGrath et al., 2014), early parental loss (Sorensen et al., 2014), second generation immigration status (Sorensen et al., 2014), low birth weight (Abel et al., 2010), short
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
gestational age (Byrne et al., 2007; Nosarti et al., 2012), obstetric complications (Cannon et al., 2002), birth order (Kemppainen et al., 2001; Pedersen and Mortensen, 2004), and Apgar score (Clarke et al., 2011). 2. Materials and methods
75
until 35 or 45, respectively. Calendar years were categorized as 1995–1999, 2000–2004 and 2005–2012. P values were based on likelihood ratio tests. We calculated the annual mortality rate (per 1000 person–years) since the date when caseness criteria were first met after adjustment for calendar time (as a time-dependent variable), age and sex.
2.1. Study population and case definitions 3. Results We used data from the Danish Civil Registration System (CRS) (Pedersen et al., 2006) which was established in 1968 to obtain information on all residents alive and living in Denmark. It includes information on personal identification number, sex, date of birth, continuously updated information on vital status, and personal identification number of parents. The personal identification number is used in all national registers enabling accurate linkage between registers. Our study population included 2,772,144 people born in Denmark in 1955–1997 and followed in the period 01.01.1995 until onset of schizophrenia, death, emigration from Denmark, or 31 December 2012. The study population and their mothers, fathers and full and half siblings were linked with the Danish Psychiatric Central Register (DPCR), (Mors et al., 2011) which was computerized in 1969. DPCR contains data on all admissions to Danish psychiatric in-patient facilities, and, from 1995, information on outpatient visits to psychiatric departments was included in the register. From 1969 to 1993, the diagnostic system used was the Danish modification of the International Classification of Diseases, 8th revision (ICD8) (WHO, 1967), and, from 1994, the International Classification of Diseases, 10th revision, Diagnostic criteria for research (ICD-10-DCR). A diagnosis of schizophrenia was defined as ICD-10-DCR F20. We compared the conventional Danish register-based case definition of schizophrenia (OCO) with the TMC case definition. OCO defines a case of schizophrenia after the first diagnosis of schizophrenia at an outpatient or inpatient contact to the secondary health care system. TMC is defined as at least 2 diagnoses with schizophrenia as inpatients. Parents and siblings were categorized hierarchically with a history of schizophrenia (ICD-8: 295; ICD-10-DCR: F20), or other mental disorders (any ICD-8 or ICD-10-DCR diagnosis), respectively, if they had been admitted to a psychiatric hospital or in outpatient care with one of these diagnoses. 2.2. Risk factors Statistics Denmark provides data on municipalities in Denmark. We used a classification according to degree of urbanization (capital, capital suburb, provincial city, provincial town, or rural areas). Maternal and paternal age at birth was subdivided into 5 and 6 categories, respectively. Information on parental loss was obtained from the Danish Civil Registration System. Immigration status was defined as the number of parents born outside Denmark (0, 1, and 2). Second generation immigration status was defined as previously (Cantor-Graae and Pedersen, 2007). Apgar score, birth defects, gestational age, birth order and birth weight were obtained from the Medical Birth Registry. Information on maternal and paternal income, highest completed level of education and occupation the year the child turned 15 was obtained from Statistics Denmark. 2.3. Statistical analyses The hazard ratio (HR) of schizophrenia according to both case definitions was estimated by Cox regression. All estimates are adjusted for calendar time as a time-dependent variable, for age in the nonparametric part of the Cox model, and for sex by means of separate underlying hazard functions. Calendar year, history of mental illness in parents, siblings and half-siblings were treated as time-dependent variables, whereas all other variables were considered time independent. Parental age was categorized as 12–19, and in 5-year intervals from age 20 and
The OCO case definition identified 15,074 individuals as cases and the TMC case definition captured 7562 (50%) of these cases. As shown in Table 1, TMC was associated with a higher mortality after adjustment for covariates including age and sex. The age and sex-adjusted annual mortality rate was significantly higher 1 year after the date at which the schizophrenia definition was met. Furthermore, the mortality rate continued to be higher during a 5 year observational period. Table 2 shows the HRs (and 95% CI's) associated with a range of risk factors according to the OCO or TMC case definition. Generally, effects of all of these risk factors were uniform regardless of case definition. Stronger risk estimates of family history of schizophrenia were found for the TMC definition as compared to the OCO definition. More or less similar patterns were found for advanced paternal age, parental loss, urban birth and second-generation immigration status, birth weight, gestational age, Apgar score, birth defects, parental educational attainment and birth order.
4. Discussion The main findings were more or less uniform risk estimates for most of the studied risk factors regardless of case definition. Trends were in the direction of similar or slightly higher risk estimates for the TMC. The TMC definition included fewer cases. Given the need for very large samples in future searches for identification of genetic variation associated with schizophrenia, i.e. in studies where size matters, it may be important to incorporate these considerations about case definitions of schizophrenia in register-based research. The dramatic reduction from 15,074 cases with the OCO definition to 7562 cases with the TMC definition should be interpreted with caution. It is probably the result of a combination of selection and truncation at the end of the study period. There could be artificial reasons (truncation) for not having a second treatment contact counted in this analysis. We would expect cases that become incident near the end of the study period to have the highest probability for right truncation bias. Therefore, the overlap of only 50% of the OCO and TMC definitions is probably a minimum estimate. Although, it is plausible that the TMC definition selects for more chronicity, poorer treatment response and higher risk of premature mortality, it is difficult with the current register-based design to measure the extent to which these selection forces have been in place.
Table 1 Descriptive characteristics of cases with schizophrenia according to a definition of schizophrenia requiring “one contact only” (OCO) or at least 2 inpatient contacts with schizophrenia (TMC).
Males N (% of total) Age, years (mean and SD) at the time when caseness criteria was first met
OCO
TMC
9174 (61%) 28.87 (8.71)
4.542 (60%) 30.31 (8.73)
Annual mortality rate (per 1000 person–years) since the date when caseness criteria were first met 1st 9.59 (8.12–11.32) 16.08 (13.40–19.29) 2nd 10.07 (8.49–11.93) 10.78 (8.54–13.60) 3rd 7.36 (5.98–9.06) 11.28 (8.89–14.31) 4th 7.24 (5.82–9.02) 9.78 (7.49–12.77) 5th 7.05 (5.58–8.91) 11.40 (8.80–14.78)
76
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
Table 2 Hazard ratio (HR) for a wide range of established and possible risk factors for schizophrenia according to a definition of schizophrenia requiring “one contact only” (OCO) or at least 2 inpatient contacts with schizophrenia (TMC) of schizophrenia. Source population
OCO
(N)
Cases (N)
TMC HR (95% CI)
Cases (N)
HR (95% CI)
Maternal age at childbirth, years 12–19 20–24 25–29 30–34 35+
179,175 839,116 978,472 543,737 231,644
1299 4996 4893 2620 1266
1.47 (1.38–1.56) 1.15 (1.10–1.19) 1.00 (ref) 1.06 (1.01–1.11) 1.29 (1.22–1.38)
684 2627 2389 1235 627
1.49 (1.37–1.62) 1.20 (1.13–1.26) 1.00 (ref) 1.03 (0.96–1.10) 1.29 (1.18–1.41)
Paternal age at childbirth, years 12–19 20–24 25–29 30–34 35–44 45+
42,600 478,734 940,418 728,916 486,736 65,901
303 2975 4881 3597 2640 430
1.42 (1.27–1.60) 1.18 (1.13–1.24) 1.00 (ref) 1.02 (0.98–1.06) 1.21 (1.16–1.27) 1.53 (1.38–1.69)
166 1593 2463 1721 1277 219
1.47 (1.26–1.73) 1.22 (1.15–1.30) 1.00 (ref) 0.98 (0.92–1.04) 1.17 (1.09–1.25) 1.50 (1.31–1.73)
Maternal vital status at child age 15 Alive Emigrated/disappeared Dead
2,738,591 6395 27,158
14,782 48 244
1.00 (ref) 1.48 (1.11–1.96) 1.60 (1.41–1.81)
7394 30 138
1.00 (ref) 1.97 (1.37–2.81) 1.80 (1.52–2.13)
Paternal vital status at child age 15 Alive Emigrated/disappeared Dead
2,651,231 16,178 61,921
13,997 137 563
1.00 (ref) 1.73 (1.47–2.05) 1.66 (1.52–1.80)
6990 78 302
1.00 (ref) 2.10 (1.68–2.63) 1.78 (1.58–1.99)
430,376 294,539 350,277 918,143 778,530
3293 1814 1843 4703 3421
1.75 (1.67–1.84) 1.34 (1.26–1.42) 1.14 (1.07–1.20) 1.07 (1.02–1.11) 1.00 (ref)
1820 939 933 2386 1484
2.14 (2.00–2.29) 1.60 (1.48–1.74) 1.30 (1.20–1.41) 1.21 (1.14–1.30) 1.00 (ref)
Number of parents not born in Denmark 0 1 2
2,500,594 143,507 62,379
12,921 1186 479
1.00 (ref) 1.67 (1.57–1.77) 1.92 (1.75–2.10)
6456 618 238
1.00 (ref) 1.80 (1.66–1.96) 2.19 (1.92–2.50)
Maternal psychiatric history None Any psychiatric diagnoses Schizophrenia
2,448,001 312,593 11,550
11,983 2748 343
1.00 (ref) 2.41 (2.31–2.51) 7.15 (6.43–7.96)
5806 1556 200
1.00 (ref) 2.75 (2.60–2.91) 8.40 (7.30–9.68)
Paternal psychiatric history None Any psychiatric diagnoses Schizophrenia Father unknown
2,467,790 253,217 8323 42,814
12,301 2177 219 377
1.00 (ref) 2.26 (2.16–2.36) 5.94 (5.19–6.79) 2.39 (2.16–2.65)
6089 1155 126 192
1.00 (ref) 2.38 (2.23–2.53) 6.99 (5.86–8.34) 2.27 (1.96–2.62)
Siblings' psychiatric history None Any psychiatric diagnoses Schizophrenia No known full siblings
1,962,907 303,728 26,300 479,209
9044 1874 530 3626
1.00 (ref) 2.42 (2.30–2.54) 7.39 (6.77–8.07) 1.80 (1.73–1.87)
4383 1037 298 1844
1.00 (ref) 2.63 (2.45–2.81) 7.80 (6.93–8.78) 1.88 (1.78–1.98)
Half siblings' psychiatric history None Any psychiatric diagnoses Schizophrenia No known half siblings
485,176 151,159 13,669 2,122,140
4163 1155 156 9600
1.00 (ref) 1.56 (1.46–1.67) 2.32 (1.98–2.72) 0.61 (0.59–0.63)
2081 623 97 4761
1.00 (ref) 1.61 (1.47–1.76) 2.71 (2.21–3.33) 0.58 (0.55–0.61)
Apgar score 5 min after birth 0–3 4–6 7–9 10
3846 5903 67,327 1,093,100
18 34 392 6,97
1.48 (0.93–2.35) 1.06 (0.76–1.49) 1.10 (1.00–1.22) 1.00 (ref)
6 14 176 2619
1.39 (0.63–3.11) 1.03 (0.61–1.75) 1.18 (1.02–1.38) 1.00 (ref)
Birth defects Yes Suspected No
18,371 2073 1,482,530
105 14 8919
1.01 (0.83–1.22) 1.03 (0.61–1.74) 1.00 (ref)
55 3 4230
1.13 (0.87–1.48) 0.59 (0.19–1.84) 1.00 (ref)
Gestational age, weeks 0–36 weeks 37–39 weeks 40+ weeks
68,053 386,410 2,317,681
517 2112 12,445
1.44 (1.32–1.58) 1.15 (1.10–1.21) 1.00 (ref)
252 899 6411
1.55 (1.36–1.76) 1.11 (1.03–1.20) 1.00 (ref)
Urbanicity at birth Capital Capital suburb Provincial city Provincial town Rural area
Birth weight, g
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
77
Table 2 (continued) Source population
OCO
(N)
Cases (N)
TMC HR (95% CI)
Cases (N)
HR (95% CI)
0–2500 g 2501–3000 g 3001–3500 g 3501+ g
79,271 220,222 512,413 1,945,223
642 1469 3183 9749
1.40 (1.29–1.52) 1.12 (1.06–1.19) 1.03 (0.98–1.07) 1.00 (ref)
337 732 1551 4931
1.68 (1.50–1.88) 1.27 (1.17–1.38) 1.14 (1.07–1.22) 1.00 (ref)
Birth order 1st 2nd 3rd 4th–18th
1,286,644 996,226 361,365 127,909
6890 5146 2100 938
1.00 (ref) 0.93 (0.90–0.97) 1.04 (0.99–1.10) 1.30 (1.22–1.39)
3430 2613 1052 467
1.00 (ref) 0.96 (0.91–1.01) 1.05 (0.98–1.13) 1.30 (1.18–1.43)
708,022 757,964 499,174
5094 3633 2508
1.36 (1.31–1.42) 1.00 (ref) 1.05 (1.00–1.11)
2611 1728 1217
1.38 (1.30–1.47) 1.00 (ref) 1.11 (1.03–1.19)
3853 4535 2280
550,140 907,384 441,897
1.37 (1.31–1.43) 1.00 (ref) 1.06 (1.01–1.12)
1927 2237 1105
1.34 (1.26–1.42) 1.00 (ref) 1.06 (0.99–1.14)
Maternal level of education at child age 15 Primary school High school/vocational training Higher education Paternal level of education at child age 15 years Primary school High school/vocational training Higher education
A slightly stronger effect of family history of schizophrenia was found for the TMC definition. In previous register-based studies with the rather narrow concept of (ICD-8) schizophrenia, rather strong effects of family history were also noted (Munk-Jorgensen and Mortensen, 1997; Sorensen et al., 2004). For the other well-established risk factors, we observed practically no dilution of the effects when comparing the OCO and TMC definitions. The observations for low birth weight and short gestational age corroborate a large body of existing evidence suggesting that suboptimal prenatal growth and postnatal growth predict risk of schizophrenia (Wahlbeck et al., 2001; Nilsson et al., 2005; Byrne et al., 2007; Nielsen et al., 2013). Choice of case definition also had little impact on our findings that a high (N 4) birth order, or a low parental education was associated with increased risk. There are several possible reasons why the number of included cases was reduced by almost 50% with the TMC vs. the OCO case definition. Some patients were probably subsequently readmitted, but assigned other psychiatric diagnoses; and some were never readmitted for a variety of reasons, including premature mortality. Thirdly, some patients with a diagnosis of schizophrenia probably had stabilized or improved so as not to require a second inpatient schizophrenia diagnosis. A substantial proportion of patients with schizophrenia do not require subsequent psychiatric admissions despite a low level of functioning and they are typically referred to sheltered institutions. Our interpretation of the findings of higher overall mortality of the TMC (1 and 5 years after the schizophrenia definition was met) is that more severe cases and possibly with more comorbid diagnoses could have been selected for.
4.1. Strengths and limitations The primary strength of this study is the prospective design, the population-based nationwide registers, and the large sample. Exposures were recorded independently of the outcome and were not subject to recall bias. Our analytic approach precluded emigration bias. Obviously, examining mortality as a function of choice of case definition of schizophrenia is a rather crude measure. This could be refined by using other and more appropriate measures of outcome in future studies. Limitations of the study were that case definitions of schizophrenia could not be directly assessed towards a gold standard. Therefore we cannot exclude that the OCO definition may have included more false positive diagnoses of schizophrenia.
4.2. Future directions Examining the epidemiology of schizophrenia often requires very large samples. Here, population-based registers (not only from Denmark and Sweden but from several other countries as well) could probably be used and combined more. In 2010, combined Swedish and Danish data examined the relationship between birth weight and risk of schizophrenia (Abel et al., 2010). In this register-based study, a common case definition was used. The Danish register included out-patient data since year 1995 and the Swedish register since year 2002. Differences between the Swedish Patient Register and the Danish Psychiatric Register were handled in various manners including stratification by country. Other register-based studies used sensitivity analysis to examine effects of different outcome definitions (Sorensen et al., 2009; Giordano et al., 2014). Giordano et al. recently used Swedish national registry databases and a co-relative case–control design with full-sibling, half-sibling and first-cousin comparisons, alongside a general Swedish population sample to examine effects of a diagnosis of cannabis abuse on the risk of schizophrenia. After requiring two separate registrations of cannabis abuse prior to two separate later diagnoses of schizophrenia, the observed association between cannabis abuse and schizophrenia risk increased modestly in the general population (Giordano et al., 2014). Although our study suggests that trends were in the direction of slightly higher risk estimates for the TMC, the introduction of a “correction factor” for inter-country population cohort studies would seem less appealing. A procedure used previously (Abel et al., 2010) along with sensitivity analyses might be preferable. In future studies, the potential impact of different case definitions of schizophrenia for risk scores (often the sum of dichotomized variables, for instance presence or absence of a specific obstetric complication diagnosis) could be explored. Variation and uncertainty introduced by different caseness definitions may accumulate with increasing complexity of the risk score. The development of a risk score usually is based on a discovery sample while external validation of the score should take place in an independent replication sample. Here, the issue of when to use alternative case definitions may be relevant too. Role of funding sources The study was supported financially by: grants from The Lundbeck Foundation and The Stanley Foundation. Neither funder had a role in the design and conduct of the study, collection, management, analyses, and interpretation of the data, and preparation, review, or approval of the manuscript and the decision to submit the manuscript for publication.
78
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
Contributors Ole Mors and PB Mortensen contributed to the design of the project. JT. Larsen and L. Petersen had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. H. Sørensen wrote the initial draft and M. Nordentoft contributed to the completion of the manuscript together with all of the authors listed. Conflicts of interest None. Acknowledgment The research leading to these results has received funding from The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark.
References Abel, K.M., Wicks, S., Susser, E.S., Dalman, C., Pedersen, M.G., Mortensen, P.B., Webb, R.T., 2010. Birth weight, schizophrenia, and adult mental disorder: is risk confined to the smallest babies? Arch. Gen. Psychiatry 67, 923–930. Byrne, M., Agerbo, E., Bennedsen, B., Eaton, W.W., Mortensen, P.B., 2007. Obstetric conditions and risk of first admission with schizophrenia: a Danish national register based study. Schizophr. Res. 97, 51–59. Cannon, M., Jones, P.B., Murray, R.M., 2002. Obstetric complications and schizophrenia: historical and meta-analytic review. Am. J. Psychiatry 159, 1080–1092. Cantor-Graae, E., Pedersen, C.B., 2007. Risk of schizophrenia in second-generation immigrants: a Danish population-based cohort study. Psychol. Med. 37, 485–494. Clarke, M.C., Tanskanen, A., Huttunen, M., Leon, D.A., Murray, R.M., Jones, P.B., Cannon, M., 2011. Increased risk of schizophrenia from additive interaction between infant motor developmental delay and obstetric complications: evidence from a population-based longitudinal study. Am. J. Psychiatry 168, 1295–1302. Dalman, C., Broms, J., Cullberg, J., Allebeck, P., 2002. Young cases of schizophrenia identified in a national inpatient register — are the diagnoses valid? Soc. Psychiatry Psychiatr. Epidemiol. 37, 527–531. Ekholm, B., Ekholm, A., Adolfsson, R., Vares, M., Osby, U., Sedvall, G.C., Jonsson, E.G., 2005. Evaluation of diagnostic procedures in Swedish patients with schizophrenia and related psychoses. Nord. J. Psychiatry 59, 457–464. Fazel, S., Langstrom, N., Hjern, A., Grann, M., Lichtenstein, P., 2009. Schizophrenia, substance abuse, and violent crime. JAMA 301, 2016–2023. Giordano, G.N., Ohlsson, H., Sundquist, K., Sundquist, J., Kendler, K.S., 2014. The association between cannabis abuse and subsequent schizophrenia: a Swedish national co-relative control study. Psychol. Med. 1–8. Kemppainen, L., Veijola, J., Jokelainen, J., Hartikainen, A.L., Jarvelin, M.R., Jones, P., Croudace, T., Isohanni, M., 2001. Birth order and risk for schizophrenia: a 31-year follow-up of the Northern Finland 1966 Birth Cohort. Acta Psychiatr. Scand. 104, 148–152. Lichtenstein, P., Bjork, C., Hultman, C.M., Scolnick, E., Sklar, P., Sullivan, P.F., 2006. Recurrence risks for schizophrenia in a Swedish national cohort. Psychol. Med. 36, 1417–1425. McGrath, J.J., Petersen, L., Agerbo, E., Mors, O., Mortensen, P.B., Pedersen, C.B., 2014. A comprehensive assessment of parental age and psychiatric disorders. JAMA Psychiatry 71, 301–309. Mors, O., Perto, G.P., Mortensen, P.B., 2011. The Danish Psychiatric Central Research Register. Scand. J. Public Health 39, 54–57. Mortensen, P.B., Pedersen, C.B., Westergaard, T., Wohlfahrt, J., Ewald, H., Mors, O., Andersen, P.K., Melbye, M., 1999. Effects of family history and place and season of birth on the risk of schizophrenia. N. Engl. J. Med. 340, 603–608. Munk-Jorgensen, P., Mortensen, P.B., 1997. The Danish Psychiatric Central Register. Dan. Med. Bull. 44, 82–84. Nielsen, P.R., Mortensen, P.B., Dalman, C., Henriksen, T.B., Pedersen, M.G., Pedersen, C.B., Agerbo, E., 2013. Fetal growth and schizophrenia: a nested case–control and case– sibling study. Schizophr. Bull. 39, 1337–1342. Nilsson, E., Stalberg, G., Lichtenstein, P., Cnattingius, S., Olausson, P.O., Hultman, C.M., 2005. Fetal growth restriction and schizophrenia: a Swedish twin study. Twin Res. Hum. Genet. 8, 402–408. Nosarti, C., Reichenberg, A., Murray, R.M., Cnattingius, S., Lambe, M.P., Yin, L., MacCabe, J., Rifkin, L., Hultman, C.M., 2012. Preterm birth and psychiatric disorders in young adult life. Arch. Gen. Psychiatry 69, E1–E8. Pedersen, C.B., Mortensen, P.B., 2001. Evidence of a dose–response relationship between urbanicity during upbringing and schizophrenia risk. Arch. Gen. Psychiatry 58, 1039–1046. Pedersen, C.B., Mortensen, P.B., 2004. Sibship characteristics during upbringing and schizophrenia risk. Am. J. Epidemiol. 160, 652–660. Pedersen, C.B., Gotzsche, H., Moller, J.O., Mortensen, P.B., 2006. The Danish Civil Registration System. A cohort of eight million persons. Dan. Med. Bull. 53, 441–449.
Purcell, S.M., Moran, J.L., Fromer, M., Ruderfer, D., Solovieff, N., Roussos, P., O'Dushlaine, C., Chambert, K., Bergen, S.E., Kahler, A., Duncan, L., Stahl, E., Genovese, G., Fernandez, E., Collins, M.O., Komiyama, N.H., Choudhary, J.S., Magnusson, P.K., Banks, E., Shakir, K., Garimella, K., Fennell, T., DePristo, M., Grant, S.G., Haggarty, S.J., Gabriel, S., Scolnick, E.M., Lander, E.S., Hultman, C.M., Sullivan, P.F., McCarroll, S.A., Sklar, P., 2014. A polygenic burden of rare disruptive mutations in schizophrenia. Nature 506, 185–190. Rees, E., Walters, J.T., Chambert, K.D., O'Dushlaine, C., Szatkiewicz, J., Richards, A.L., Georgieva, L., Mahoney-Davies, G., Legge, S.E., Moran, J.L., Genovese, G., Levinson, D., Morris, D.W., Cormican, P., Kendler, K.S., O'Neill, F.A., Riley, B., Gill, M., Corvin, A., Sklar, P., Hultman, C., Pato, C., Pato, M., Sullivan, P.F., Gejman, P.V., McCarroll, S.A., O'Donovan, M.C., Owen, M.J., Kirov, G., 2014. CNV analysis in a large schizophrenia sample implicates deletions at 16p12.1 and SLC1A1 and duplications at 1p36.33 and CGNL1. Hum. Mol. Genet. 23, 1669–1676. Ripke, S., O'Dushlaine, C., Chambert, K., Moran, J.L., Kahler, A.K., Akterin, S., Bergen, S.E., Collins, A.L., Crowley, J.J., Fromer, M., Kim, Y., Lee, S.H., Magnusson, P.K., Sanchez, N., Stahl, E.A., Williams, S., Wray, N.R., Xia, K., Bettella, F., Borglum, A.D., Bulik-Sullivan, B.K., Cormican, P., Craddock, N., de, L.C., Durmishi, N., Gill, M., Golimbet, V., Hamshere, M.L., Holmans, P., Hougaard, D.M., Kendler, K.S., Lin, K., Morris, D.W., Mors, O., Mortensen, P.B., Neale, B.M., O'Neill,, F.A., Owen, M.J., Milovancevic, M.P., Posthuma, D., Powell, J., Richards, A.L., Riley, B.P., Ruderfer, D., Rujescu, D., Sigurdsson, E., Silagadze, T., Smit, A.B., Stefansson, H., Steinberg, S., Suvisaari, J., Tosato, S., Verhage, M., Walters, J.T., Levinson, D.F., Gejman, P.V., Kendler, K.S., Laurent, C., Mowry, B.J., O'Donovan, M.C., Owen, M.J., Pulver, A.E., Riley, B.P., Schwab, S.G., Wildenauer, D.B., Dudbridge, F., Holmans, P., Shi, J., Albus, M., Alexander, M., Campion, D., Cohen, D., Dikeos, D., Duan, J., Eichhammer, P., Godard, S., Hansen, M., Lerer, F.B., Liang, K.Y., Maier, W., Mallet, J., Nertney, D.A., Nestadt, G., Norton, N., O'Neill, F.A., Papadimitriou, G.N., Ribble, R., Sanders, A.R., Silverman, J.M., Walsh, D., Williams, N.M., Wormley, B., Arranz, M.J., Bakker, S., Bender, S., Bramon, E., Collier, D., Crespo-Facorro, B., Hall, J., Iyegbe, C., Jablensky, A., Kahn, R.S., Kalaydjieva, L., Lawrie, S., Lewis, C.M., Lin, K., Linszen, D.H., Mata, I., McIntosh, A., Murray, R.M., Ophoff, R.A., Powell, J., Rujescu, D., Van, O.J., Walshe, M., Weisbrod, M., Wiersma, D., Donnelly, P., Barroso, I., Blackwell, J.M., Bramon, E., Brown, M.A., Casas, J.P., Corvin, A.P., Deloukas, P., Duncanson, A., Jankowski, J., Markus, H.S., Mathew, C.G., Palmer, C.N., Plomin, R., Rautanen, A., Sawcer, S.J., Trembath, R.C., Viswanathan, A.C., Wood, N.W., Spencer, C.C., Band, G., Bellenguez, C., Freeman, C., Hellenthal, G., Giannoulatou, E., Pirinen, M., Pearson, R.D., Strange, A., Su, Z., Vukcevic, D., Donnelly, P., Langford, C., Hunt, S.E., Edkins, S., Gwilliam, R., Blackburn, H., Bumpstead, S.J., Dronov, S., Gillman, M., Gray, E., Hammond, N., Jayakumar, A., McCann, O.T., Liddle, J., Potter, S.C., Ravindrarajah, R., Ricketts, M., TashakkoriGhanbaria, A., Waller, M.J., Weston, P., Widaa, S., Whittaker, P., Barroso, I., Deloukas, P., Mathew, C.G., Blackwell, J.M., Brown, M.A., Corvin, A.P., McCarthy, M.I., Spencer, C.C., Bramon, E., Corvin, A.P., O'Donovan, M.C., Stefansson, K., Scolnick, E., Purcell, S., McCarroll, S.A., Sklar, P., Hultman, C.M., Sullivan, P.F., 2013. Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nat. Genet. 45, 1150–1159. Robins, E., Guze, S.B., 1970. Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am. J. Psychiatry 126, 983–987. Ruderfer, D.M., Fanous, A.H., Ripke, S., Mcquillin, A., Amdur, R.L., Gejman, P.V., O'Donovan, M.C., Andreassen, O.A., Djurovic, S., Hultman, C.M., Kelsoe, J.R., Jamain, S., Landen, M., Leboyer, M., Nimgaonkar, V., Nurnberger, J., Smoller, J.W., Craddock, N., Corvin, A., Sullivan, P.F., Holmans, P., Sklar, P., Kendler, K.S., 2014. Polygenic dissection of diagnosis and clinical dimensions of bipolar disorder and schizophrenia. Mol. Psychiatry 19 (9), 1017–1024. Sorensen, H.J., Mortensen, E.L., Reinisch, J.M., Mednick, S.A., 2004. Association between prenatal exposure to analgesics and risk of schizophrenia. Br. J. Psychiatry 185, 366–371. Sorensen, H.J., Mortensen, E.L., Reinisch, J.M., Mednick, S.A., 2009. Association between prenatal exposure to bacterial infection and risk of schizophrenia. Schizophr. Bull. 35, 631–637. Sorensen, H.J., Nielsen, P.R., Pedersen, C.B., Benros, M.E., Nordentoft, M., Mortensen, P.B., 2014. Population impact of familial and environmental risk factors for schizophrenia: a nationwide study. Schizophr. Res. 153 (1-3), 214–219. Szatkiewicz, J.P., O'Dushlaine, C., Chen, G., Chambert, K., Moran, J.L., Neale, B.M., Fromer, M., Ruderfer, D., Akterin, S., Bergen, S.E., Kahler, A., Magnusson, P.K., Kim, Y., Crowley, J.J., Rees, E., Kirov, G., O'Donovan, M.C., Owen, M.J., Walters, J., Scolnick, E., Sklar, P., Purcell, S., Hultman, C.M., McCarroll, S.A., Sullivan, P.F., 2014. Copy number variation in schizophrenia in Sweden. Mol. Psychiatry 19 (7), 762–773. Uggerby, P., Ostergaard, S.D., Roge, R., Correll, C.U., Nielsen, J., 2013. The validity of the schizophrenia diagnosis in the Danish Psychiatric Central Research Register is good. Dan. Med. J. 60, A4578. WHO, 1967. International Statistical Classification of Diseases, Injuries, and Causes of Death. Eight Revision (1965) Geneva. Wahlbeck, K., Forsen, T., Osmond, C., Barker, D.J., Eriksson, J.G., 2001. Association of schizophrenia with low maternal body mass index, small size at birth, and thinness during childhood. Arch. Gen. Psychiatry 58, 48–52.
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Analysis of risk factors for schizophrenia with two different case definitions: A nationwide register-based external validation study☆ Holger J. Sørensen a,e,⁎, Janne T. Larsen b,e, Ole Mors d,e, Merete Nordentoft a,e, Preben B. Mortensen b,c,e, Liselotte Petersen b,e a
Psychiatric Centre Copenhagen, Capital Region of Denmark, University Hospital of Copenhagen, Bispebjerg Bakke 23, 2400 NV, Denmark National Centre for Register-based Research, Aarhus University, Business and Social Sciences, Aarhus, Denmark Centre for Integrated Register-based Research, CIRRAU, Aarhus University, Aarhus, Denmark d Research Department P, Aarhus University Hospital, Risskov, Denmark e The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark b c
a r t i c l e
i n f o
Article history: Received 9 October 2014 Received in revised form 29 December 2014 Accepted 8 January 2015 Available online 23 January 2015 Keywords: Schizophrenia Case definition External validation Register-based studies
a b s t r a c t Different case definitions of schizophrenia have been used in register based research. However, no previous study has externally validated two different case definitions of schizophrenia against a wide range of risk factors for schizophrenia. We investigated hazard ratios (HRs) for a wide range of risk factors for ICD-10 DCR schizophrenia using a nationwide Danish sample of 2,772,144 residents born in 1955–1997. We compared one contact only (OCO) (the case definition of schizophrenia used in Danish register based studies) with two or more contacts (TMC) (a case definition of at least 2 inpatient contacts with schizophrenia). During the follow-up, the OCO definition included 15,074 and the TMC 7562 cases; i.e. half as many. The TMC case definition appeared to select for a worse illness course. A wide range of risk factors were uniformly associated with both case definitions and only slightly higher risk estimates were found for the TMC definition. Choosing at least 2 inpatient contacts with schizophrenia (TMC) instead of the currently used case definition would result in almost similar risk estimates for many well-established risk factors. However, this would also introduce selection and include considerably fewer cases and reduce power of e.g. genetic studies based on register-diagnosed cases only. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The decision on when to define a case of schizophrenia in registerbased research is guided by several considerations. In Denmark, one contact to the secondary health care system with schizophrenia has been viewed as sufficient to define a case in register-based research. To increase the diagnostic validity, by minimizing false-positive diagnoses, researchers using the Swedish hospital discharge register sometimes employ a definition requiring a diagnosis of schizophrenia on two separate treatment inpatient contacts (Lichtenstein et al., 2006). However, a case definition of schizophrenia demanding two or more contacts (TMC) could lead to increased risk of selection. On the other hand, the advantage of the TMC definition is reduced heterogeneity among cases and a more homogeneous sample suited for e.g. biological studies (Fazel et al., 2009; Ripke et al., 2013; Ruderfer et al., 2014;
☆ Presented at The 4th Biennial Schizophrenia International Research Society Conference, 5–9 April 2014 in Florence, Italy. ⁎ Corresponding author at: Mental Health Centre Copenhagen, Bispebjerg Bakke 23, 13A, DK 2400 NV, Copenhagen, Denmark. E-mail address: [email protected] (H.J. Sørensen).
http://dx.doi.org/10.1016/j.schres.2015.01.018 0920-9964/© 2015 Elsevier B.V. All rights reserved.
Purcell et al., 2014; Rees et al., 2014; Szatkiewicz et al., 2014). Conversely, the potential disadvantage of the OCO (one case only) definition used in Denmark is a higher risk of false-positive diagnoses. There has been several reports (Dalman et al., 2002; Ekholm et al., 2005; Uggerby et al., 2013) suggesting that the validity of a schizophrenia diagnosis is good in Swedish as well as in Danish nationwide registers. However, since there are no supportive diagnostic tests for schizophrenia, it may be important, when designing e.g. future genetic studies based on register-based samples only, to consider when to use alternative ascertainment or not. In large register-based studies, sample size and effects of common and well-established risk factors could potentially depend on the case definition of schizophrenia. To study this further, we here examine the TMC definition versus the OCO definition in an external validation study. We decided to use well-established risk factors for schizophrenia as external validators as suggested previously (Robins and Guze, 1970). These included family history of mental illness (Mortensen et al., 1999), urban place of birth (Pedersen and Mortensen, 2001), unknown parent status (Mortensen et al., 1999), advanced paternal age (Sorensen et al., 2014; McGrath et al., 2014), early parental loss (Sorensen et al., 2014), second generation immigration status (Sorensen et al., 2014), low birth weight (Abel et al., 2010), short
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
gestational age (Byrne et al., 2007; Nosarti et al., 2012), obstetric complications (Cannon et al., 2002), birth order (Kemppainen et al., 2001; Pedersen and Mortensen, 2004), and Apgar score (Clarke et al., 2011). 2. Materials and methods
75
until 35 or 45, respectively. Calendar years were categorized as 1995–1999, 2000–2004 and 2005–2012. P values were based on likelihood ratio tests. We calculated the annual mortality rate (per 1000 person–years) since the date when caseness criteria were first met after adjustment for calendar time (as a time-dependent variable), age and sex.
2.1. Study population and case definitions 3. Results We used data from the Danish Civil Registration System (CRS) (Pedersen et al., 2006) which was established in 1968 to obtain information on all residents alive and living in Denmark. It includes information on personal identification number, sex, date of birth, continuously updated information on vital status, and personal identification number of parents. The personal identification number is used in all national registers enabling accurate linkage between registers. Our study population included 2,772,144 people born in Denmark in 1955–1997 and followed in the period 01.01.1995 until onset of schizophrenia, death, emigration from Denmark, or 31 December 2012. The study population and their mothers, fathers and full and half siblings were linked with the Danish Psychiatric Central Register (DPCR), (Mors et al., 2011) which was computerized in 1969. DPCR contains data on all admissions to Danish psychiatric in-patient facilities, and, from 1995, information on outpatient visits to psychiatric departments was included in the register. From 1969 to 1993, the diagnostic system used was the Danish modification of the International Classification of Diseases, 8th revision (ICD8) (WHO, 1967), and, from 1994, the International Classification of Diseases, 10th revision, Diagnostic criteria for research (ICD-10-DCR). A diagnosis of schizophrenia was defined as ICD-10-DCR F20. We compared the conventional Danish register-based case definition of schizophrenia (OCO) with the TMC case definition. OCO defines a case of schizophrenia after the first diagnosis of schizophrenia at an outpatient or inpatient contact to the secondary health care system. TMC is defined as at least 2 diagnoses with schizophrenia as inpatients. Parents and siblings were categorized hierarchically with a history of schizophrenia (ICD-8: 295; ICD-10-DCR: F20), or other mental disorders (any ICD-8 or ICD-10-DCR diagnosis), respectively, if they had been admitted to a psychiatric hospital or in outpatient care with one of these diagnoses. 2.2. Risk factors Statistics Denmark provides data on municipalities in Denmark. We used a classification according to degree of urbanization (capital, capital suburb, provincial city, provincial town, or rural areas). Maternal and paternal age at birth was subdivided into 5 and 6 categories, respectively. Information on parental loss was obtained from the Danish Civil Registration System. Immigration status was defined as the number of parents born outside Denmark (0, 1, and 2). Second generation immigration status was defined as previously (Cantor-Graae and Pedersen, 2007). Apgar score, birth defects, gestational age, birth order and birth weight were obtained from the Medical Birth Registry. Information on maternal and paternal income, highest completed level of education and occupation the year the child turned 15 was obtained from Statistics Denmark. 2.3. Statistical analyses The hazard ratio (HR) of schizophrenia according to both case definitions was estimated by Cox regression. All estimates are adjusted for calendar time as a time-dependent variable, for age in the nonparametric part of the Cox model, and for sex by means of separate underlying hazard functions. Calendar year, history of mental illness in parents, siblings and half-siblings were treated as time-dependent variables, whereas all other variables were considered time independent. Parental age was categorized as 12–19, and in 5-year intervals from age 20 and
The OCO case definition identified 15,074 individuals as cases and the TMC case definition captured 7562 (50%) of these cases. As shown in Table 1, TMC was associated with a higher mortality after adjustment for covariates including age and sex. The age and sex-adjusted annual mortality rate was significantly higher 1 year after the date at which the schizophrenia definition was met. Furthermore, the mortality rate continued to be higher during a 5 year observational period. Table 2 shows the HRs (and 95% CI's) associated with a range of risk factors according to the OCO or TMC case definition. Generally, effects of all of these risk factors were uniform regardless of case definition. Stronger risk estimates of family history of schizophrenia were found for the TMC definition as compared to the OCO definition. More or less similar patterns were found for advanced paternal age, parental loss, urban birth and second-generation immigration status, birth weight, gestational age, Apgar score, birth defects, parental educational attainment and birth order.
4. Discussion The main findings were more or less uniform risk estimates for most of the studied risk factors regardless of case definition. Trends were in the direction of similar or slightly higher risk estimates for the TMC. The TMC definition included fewer cases. Given the need for very large samples in future searches for identification of genetic variation associated with schizophrenia, i.e. in studies where size matters, it may be important to incorporate these considerations about case definitions of schizophrenia in register-based research. The dramatic reduction from 15,074 cases with the OCO definition to 7562 cases with the TMC definition should be interpreted with caution. It is probably the result of a combination of selection and truncation at the end of the study period. There could be artificial reasons (truncation) for not having a second treatment contact counted in this analysis. We would expect cases that become incident near the end of the study period to have the highest probability for right truncation bias. Therefore, the overlap of only 50% of the OCO and TMC definitions is probably a minimum estimate. Although, it is plausible that the TMC definition selects for more chronicity, poorer treatment response and higher risk of premature mortality, it is difficult with the current register-based design to measure the extent to which these selection forces have been in place.
Table 1 Descriptive characteristics of cases with schizophrenia according to a definition of schizophrenia requiring “one contact only” (OCO) or at least 2 inpatient contacts with schizophrenia (TMC).
Males N (% of total) Age, years (mean and SD) at the time when caseness criteria was first met
OCO
TMC
9174 (61%) 28.87 (8.71)
4.542 (60%) 30.31 (8.73)
Annual mortality rate (per 1000 person–years) since the date when caseness criteria were first met 1st 9.59 (8.12–11.32) 16.08 (13.40–19.29) 2nd 10.07 (8.49–11.93) 10.78 (8.54–13.60) 3rd 7.36 (5.98–9.06) 11.28 (8.89–14.31) 4th 7.24 (5.82–9.02) 9.78 (7.49–12.77) 5th 7.05 (5.58–8.91) 11.40 (8.80–14.78)
76
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
Table 2 Hazard ratio (HR) for a wide range of established and possible risk factors for schizophrenia according to a definition of schizophrenia requiring “one contact only” (OCO) or at least 2 inpatient contacts with schizophrenia (TMC) of schizophrenia. Source population
OCO
(N)
Cases (N)
TMC HR (95% CI)
Cases (N)
HR (95% CI)
Maternal age at childbirth, years 12–19 20–24 25–29 30–34 35+
179,175 839,116 978,472 543,737 231,644
1299 4996 4893 2620 1266
1.47 (1.38–1.56) 1.15 (1.10–1.19) 1.00 (ref) 1.06 (1.01–1.11) 1.29 (1.22–1.38)
684 2627 2389 1235 627
1.49 (1.37–1.62) 1.20 (1.13–1.26) 1.00 (ref) 1.03 (0.96–1.10) 1.29 (1.18–1.41)
Paternal age at childbirth, years 12–19 20–24 25–29 30–34 35–44 45+
42,600 478,734 940,418 728,916 486,736 65,901
303 2975 4881 3597 2640 430
1.42 (1.27–1.60) 1.18 (1.13–1.24) 1.00 (ref) 1.02 (0.98–1.06) 1.21 (1.16–1.27) 1.53 (1.38–1.69)
166 1593 2463 1721 1277 219
1.47 (1.26–1.73) 1.22 (1.15–1.30) 1.00 (ref) 0.98 (0.92–1.04) 1.17 (1.09–1.25) 1.50 (1.31–1.73)
Maternal vital status at child age 15 Alive Emigrated/disappeared Dead
2,738,591 6395 27,158
14,782 48 244
1.00 (ref) 1.48 (1.11–1.96) 1.60 (1.41–1.81)
7394 30 138
1.00 (ref) 1.97 (1.37–2.81) 1.80 (1.52–2.13)
Paternal vital status at child age 15 Alive Emigrated/disappeared Dead
2,651,231 16,178 61,921
13,997 137 563
1.00 (ref) 1.73 (1.47–2.05) 1.66 (1.52–1.80)
6990 78 302
1.00 (ref) 2.10 (1.68–2.63) 1.78 (1.58–1.99)
430,376 294,539 350,277 918,143 778,530
3293 1814 1843 4703 3421
1.75 (1.67–1.84) 1.34 (1.26–1.42) 1.14 (1.07–1.20) 1.07 (1.02–1.11) 1.00 (ref)
1820 939 933 2386 1484
2.14 (2.00–2.29) 1.60 (1.48–1.74) 1.30 (1.20–1.41) 1.21 (1.14–1.30) 1.00 (ref)
Number of parents not born in Denmark 0 1 2
2,500,594 143,507 62,379
12,921 1186 479
1.00 (ref) 1.67 (1.57–1.77) 1.92 (1.75–2.10)
6456 618 238
1.00 (ref) 1.80 (1.66–1.96) 2.19 (1.92–2.50)
Maternal psychiatric history None Any psychiatric diagnoses Schizophrenia
2,448,001 312,593 11,550
11,983 2748 343
1.00 (ref) 2.41 (2.31–2.51) 7.15 (6.43–7.96)
5806 1556 200
1.00 (ref) 2.75 (2.60–2.91) 8.40 (7.30–9.68)
Paternal psychiatric history None Any psychiatric diagnoses Schizophrenia Father unknown
2,467,790 253,217 8323 42,814
12,301 2177 219 377
1.00 (ref) 2.26 (2.16–2.36) 5.94 (5.19–6.79) 2.39 (2.16–2.65)
6089 1155 126 192
1.00 (ref) 2.38 (2.23–2.53) 6.99 (5.86–8.34) 2.27 (1.96–2.62)
Siblings' psychiatric history None Any psychiatric diagnoses Schizophrenia No known full siblings
1,962,907 303,728 26,300 479,209
9044 1874 530 3626
1.00 (ref) 2.42 (2.30–2.54) 7.39 (6.77–8.07) 1.80 (1.73–1.87)
4383 1037 298 1844
1.00 (ref) 2.63 (2.45–2.81) 7.80 (6.93–8.78) 1.88 (1.78–1.98)
Half siblings' psychiatric history None Any psychiatric diagnoses Schizophrenia No known half siblings
485,176 151,159 13,669 2,122,140
4163 1155 156 9600
1.00 (ref) 1.56 (1.46–1.67) 2.32 (1.98–2.72) 0.61 (0.59–0.63)
2081 623 97 4761
1.00 (ref) 1.61 (1.47–1.76) 2.71 (2.21–3.33) 0.58 (0.55–0.61)
Apgar score 5 min after birth 0–3 4–6 7–9 10
3846 5903 67,327 1,093,100
18 34 392 6,97
1.48 (0.93–2.35) 1.06 (0.76–1.49) 1.10 (1.00–1.22) 1.00 (ref)
6 14 176 2619
1.39 (0.63–3.11) 1.03 (0.61–1.75) 1.18 (1.02–1.38) 1.00 (ref)
Birth defects Yes Suspected No
18,371 2073 1,482,530
105 14 8919
1.01 (0.83–1.22) 1.03 (0.61–1.74) 1.00 (ref)
55 3 4230
1.13 (0.87–1.48) 0.59 (0.19–1.84) 1.00 (ref)
Gestational age, weeks 0–36 weeks 37–39 weeks 40+ weeks
68,053 386,410 2,317,681
517 2112 12,445
1.44 (1.32–1.58) 1.15 (1.10–1.21) 1.00 (ref)
252 899 6411
1.55 (1.36–1.76) 1.11 (1.03–1.20) 1.00 (ref)
Urbanicity at birth Capital Capital suburb Provincial city Provincial town Rural area
Birth weight, g
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
77
Table 2 (continued) Source population
OCO
(N)
Cases (N)
TMC HR (95% CI)
Cases (N)
HR (95% CI)
0–2500 g 2501–3000 g 3001–3500 g 3501+ g
79,271 220,222 512,413 1,945,223
642 1469 3183 9749
1.40 (1.29–1.52) 1.12 (1.06–1.19) 1.03 (0.98–1.07) 1.00 (ref)
337 732 1551 4931
1.68 (1.50–1.88) 1.27 (1.17–1.38) 1.14 (1.07–1.22) 1.00 (ref)
Birth order 1st 2nd 3rd 4th–18th
1,286,644 996,226 361,365 127,909
6890 5146 2100 938
1.00 (ref) 0.93 (0.90–0.97) 1.04 (0.99–1.10) 1.30 (1.22–1.39)
3430 2613 1052 467
1.00 (ref) 0.96 (0.91–1.01) 1.05 (0.98–1.13) 1.30 (1.18–1.43)
708,022 757,964 499,174
5094 3633 2508
1.36 (1.31–1.42) 1.00 (ref) 1.05 (1.00–1.11)
2611 1728 1217
1.38 (1.30–1.47) 1.00 (ref) 1.11 (1.03–1.19)
3853 4535 2280
550,140 907,384 441,897
1.37 (1.31–1.43) 1.00 (ref) 1.06 (1.01–1.12)
1927 2237 1105
1.34 (1.26–1.42) 1.00 (ref) 1.06 (0.99–1.14)
Maternal level of education at child age 15 Primary school High school/vocational training Higher education Paternal level of education at child age 15 years Primary school High school/vocational training Higher education
A slightly stronger effect of family history of schizophrenia was found for the TMC definition. In previous register-based studies with the rather narrow concept of (ICD-8) schizophrenia, rather strong effects of family history were also noted (Munk-Jorgensen and Mortensen, 1997; Sorensen et al., 2004). For the other well-established risk factors, we observed practically no dilution of the effects when comparing the OCO and TMC definitions. The observations for low birth weight and short gestational age corroborate a large body of existing evidence suggesting that suboptimal prenatal growth and postnatal growth predict risk of schizophrenia (Wahlbeck et al., 2001; Nilsson et al., 2005; Byrne et al., 2007; Nielsen et al., 2013). Choice of case definition also had little impact on our findings that a high (N 4) birth order, or a low parental education was associated with increased risk. There are several possible reasons why the number of included cases was reduced by almost 50% with the TMC vs. the OCO case definition. Some patients were probably subsequently readmitted, but assigned other psychiatric diagnoses; and some were never readmitted for a variety of reasons, including premature mortality. Thirdly, some patients with a diagnosis of schizophrenia probably had stabilized or improved so as not to require a second inpatient schizophrenia diagnosis. A substantial proportion of patients with schizophrenia do not require subsequent psychiatric admissions despite a low level of functioning and they are typically referred to sheltered institutions. Our interpretation of the findings of higher overall mortality of the TMC (1 and 5 years after the schizophrenia definition was met) is that more severe cases and possibly with more comorbid diagnoses could have been selected for.
4.1. Strengths and limitations The primary strength of this study is the prospective design, the population-based nationwide registers, and the large sample. Exposures were recorded independently of the outcome and were not subject to recall bias. Our analytic approach precluded emigration bias. Obviously, examining mortality as a function of choice of case definition of schizophrenia is a rather crude measure. This could be refined by using other and more appropriate measures of outcome in future studies. Limitations of the study were that case definitions of schizophrenia could not be directly assessed towards a gold standard. Therefore we cannot exclude that the OCO definition may have included more false positive diagnoses of schizophrenia.
4.2. Future directions Examining the epidemiology of schizophrenia often requires very large samples. Here, population-based registers (not only from Denmark and Sweden but from several other countries as well) could probably be used and combined more. In 2010, combined Swedish and Danish data examined the relationship between birth weight and risk of schizophrenia (Abel et al., 2010). In this register-based study, a common case definition was used. The Danish register included out-patient data since year 1995 and the Swedish register since year 2002. Differences between the Swedish Patient Register and the Danish Psychiatric Register were handled in various manners including stratification by country. Other register-based studies used sensitivity analysis to examine effects of different outcome definitions (Sorensen et al., 2009; Giordano et al., 2014). Giordano et al. recently used Swedish national registry databases and a co-relative case–control design with full-sibling, half-sibling and first-cousin comparisons, alongside a general Swedish population sample to examine effects of a diagnosis of cannabis abuse on the risk of schizophrenia. After requiring two separate registrations of cannabis abuse prior to two separate later diagnoses of schizophrenia, the observed association between cannabis abuse and schizophrenia risk increased modestly in the general population (Giordano et al., 2014). Although our study suggests that trends were in the direction of slightly higher risk estimates for the TMC, the introduction of a “correction factor” for inter-country population cohort studies would seem less appealing. A procedure used previously (Abel et al., 2010) along with sensitivity analyses might be preferable. In future studies, the potential impact of different case definitions of schizophrenia for risk scores (often the sum of dichotomized variables, for instance presence or absence of a specific obstetric complication diagnosis) could be explored. Variation and uncertainty introduced by different caseness definitions may accumulate with increasing complexity of the risk score. The development of a risk score usually is based on a discovery sample while external validation of the score should take place in an independent replication sample. Here, the issue of when to use alternative case definitions may be relevant too. Role of funding sources The study was supported financially by: grants from The Lundbeck Foundation and The Stanley Foundation. Neither funder had a role in the design and conduct of the study, collection, management, analyses, and interpretation of the data, and preparation, review, or approval of the manuscript and the decision to submit the manuscript for publication.
78
H.J. Sørensen et al. / Schizophrenia Research 162 (2015) 74–78
Contributors Ole Mors and PB Mortensen contributed to the design of the project. JT. Larsen and L. Petersen had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. H. Sørensen wrote the initial draft and M. Nordentoft contributed to the completion of the manuscript together with all of the authors listed. Conflicts of interest None. Acknowledgment The research leading to these results has received funding from The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark.
References Abel, K.M., Wicks, S., Susser, E.S., Dalman, C., Pedersen, M.G., Mortensen, P.B., Webb, R.T., 2010. Birth weight, schizophrenia, and adult mental disorder: is risk confined to the smallest babies? Arch. Gen. Psychiatry 67, 923–930. Byrne, M., Agerbo, E., Bennedsen, B., Eaton, W.W., Mortensen, P.B., 2007. Obstetric conditions and risk of first admission with schizophrenia: a Danish national register based study. Schizophr. Res. 97, 51–59. Cannon, M., Jones, P.B., Murray, R.M., 2002. Obstetric complications and schizophrenia: historical and meta-analytic review. Am. J. Psychiatry 159, 1080–1092. Cantor-Graae, E., Pedersen, C.B., 2007. Risk of schizophrenia in second-generation immigrants: a Danish population-based cohort study. Psychol. Med. 37, 485–494. Clarke, M.C., Tanskanen, A., Huttunen, M., Leon, D.A., Murray, R.M., Jones, P.B., Cannon, M., 2011. Increased risk of schizophrenia from additive interaction between infant motor developmental delay and obstetric complications: evidence from a population-based longitudinal study. Am. J. Psychiatry 168, 1295–1302. Dalman, C., Broms, J., Cullberg, J., Allebeck, P., 2002. Young cases of schizophrenia identified in a national inpatient register — are the diagnoses valid? Soc. Psychiatry Psychiatr. Epidemiol. 37, 527–531. Ekholm, B., Ekholm, A., Adolfsson, R., Vares, M., Osby, U., Sedvall, G.C., Jonsson, E.G., 2005. Evaluation of diagnostic procedures in Swedish patients with schizophrenia and related psychoses. Nord. J. Psychiatry 59, 457–464. Fazel, S., Langstrom, N., Hjern, A., Grann, M., Lichtenstein, P., 2009. Schizophrenia, substance abuse, and violent crime. JAMA 301, 2016–2023. Giordano, G.N., Ohlsson, H., Sundquist, K., Sundquist, J., Kendler, K.S., 2014. The association between cannabis abuse and subsequent schizophrenia: a Swedish national co-relative control study. Psychol. Med. 1–8. Kemppainen, L., Veijola, J., Jokelainen, J., Hartikainen, A.L., Jarvelin, M.R., Jones, P., Croudace, T., Isohanni, M., 2001. Birth order and risk for schizophrenia: a 31-year follow-up of the Northern Finland 1966 Birth Cohort. Acta Psychiatr. Scand. 104, 148–152. Lichtenstein, P., Bjork, C., Hultman, C.M., Scolnick, E., Sklar, P., Sullivan, P.F., 2006. Recurrence risks for schizophrenia in a Swedish national cohort. Psychol. Med. 36, 1417–1425. McGrath, J.J., Petersen, L., Agerbo, E., Mors, O., Mortensen, P.B., Pedersen, C.B., 2014. A comprehensive assessment of parental age and psychiatric disorders. JAMA Psychiatry 71, 301–309. Mors, O., Perto, G.P., Mortensen, P.B., 2011. The Danish Psychiatric Central Research Register. Scand. J. Public Health 39, 54–57. Mortensen, P.B., Pedersen, C.B., Westergaard, T., Wohlfahrt, J., Ewald, H., Mors, O., Andersen, P.K., Melbye, M., 1999. Effects of family history and place and season of birth on the risk of schizophrenia. N. Engl. J. Med. 340, 603–608. Munk-Jorgensen, P., Mortensen, P.B., 1997. The Danish Psychiatric Central Register. Dan. Med. Bull. 44, 82–84. Nielsen, P.R., Mortensen, P.B., Dalman, C., Henriksen, T.B., Pedersen, M.G., Pedersen, C.B., Agerbo, E., 2013. Fetal growth and schizophrenia: a nested case–control and case– sibling study. Schizophr. Bull. 39, 1337–1342. Nilsson, E., Stalberg, G., Lichtenstein, P., Cnattingius, S., Olausson, P.O., Hultman, C.M., 2005. Fetal growth restriction and schizophrenia: a Swedish twin study. Twin Res. Hum. Genet. 8, 402–408. Nosarti, C., Reichenberg, A., Murray, R.M., Cnattingius, S., Lambe, M.P., Yin, L., MacCabe, J., Rifkin, L., Hultman, C.M., 2012. Preterm birth and psychiatric disorders in young adult life. Arch. Gen. Psychiatry 69, E1–E8. Pedersen, C.B., Mortensen, P.B., 2001. Evidence of a dose–response relationship between urbanicity during upbringing and schizophrenia risk. Arch. Gen. Psychiatry 58, 1039–1046. Pedersen, C.B., Mortensen, P.B., 2004. Sibship characteristics during upbringing and schizophrenia risk. Am. J. Epidemiol. 160, 652–660. Pedersen, C.B., Gotzsche, H., Moller, J.O., Mortensen, P.B., 2006. The Danish Civil Registration System. A cohort of eight million persons. Dan. Med. Bull. 53, 441–449.
Purcell, S.M., Moran, J.L., Fromer, M., Ruderfer, D., Solovieff, N., Roussos, P., O'Dushlaine, C., Chambert, K., Bergen, S.E., Kahler, A., Duncan, L., Stahl, E., Genovese, G., Fernandez, E., Collins, M.O., Komiyama, N.H., Choudhary, J.S., Magnusson, P.K., Banks, E., Shakir, K., Garimella, K., Fennell, T., DePristo, M., Grant, S.G., Haggarty, S.J., Gabriel, S., Scolnick, E.M., Lander, E.S., Hultman, C.M., Sullivan, P.F., McCarroll, S.A., Sklar, P., 2014. A polygenic burden of rare disruptive mutations in schizophrenia. Nature 506, 185–190. Rees, E., Walters, J.T., Chambert, K.D., O'Dushlaine, C., Szatkiewicz, J., Richards, A.L., Georgieva, L., Mahoney-Davies, G., Legge, S.E., Moran, J.L., Genovese, G., Levinson, D., Morris, D.W., Cormican, P., Kendler, K.S., O'Neill, F.A., Riley, B., Gill, M., Corvin, A., Sklar, P., Hultman, C., Pato, C., Pato, M., Sullivan, P.F., Gejman, P.V., McCarroll, S.A., O'Donovan, M.C., Owen, M.J., Kirov, G., 2014. CNV analysis in a large schizophrenia sample implicates deletions at 16p12.1 and SLC1A1 and duplications at 1p36.33 and CGNL1. Hum. Mol. Genet. 23, 1669–1676. Ripke, S., O'Dushlaine, C., Chambert, K., Moran, J.L., Kahler, A.K., Akterin, S., Bergen, S.E., Collins, A.L., Crowley, J.J., Fromer, M., Kim, Y., Lee, S.H., Magnusson, P.K., Sanchez, N., Stahl, E.A., Williams, S., Wray, N.R., Xia, K., Bettella, F., Borglum, A.D., Bulik-Sullivan, B.K., Cormican, P., Craddock, N., de, L.C., Durmishi, N., Gill, M., Golimbet, V., Hamshere, M.L., Holmans, P., Hougaard, D.M., Kendler, K.S., Lin, K., Morris, D.W., Mors, O., Mortensen, P.B., Neale, B.M., O'Neill,, F.A., Owen, M.J., Milovancevic, M.P., Posthuma, D., Powell, J., Richards, A.L., Riley, B.P., Ruderfer, D., Rujescu, D., Sigurdsson, E., Silagadze, T., Smit, A.B., Stefansson, H., Steinberg, S., Suvisaari, J., Tosato, S., Verhage, M., Walters, J.T., Levinson, D.F., Gejman, P.V., Kendler, K.S., Laurent, C., Mowry, B.J., O'Donovan, M.C., Owen, M.J., Pulver, A.E., Riley, B.P., Schwab, S.G., Wildenauer, D.B., Dudbridge, F., Holmans, P., Shi, J., Albus, M., Alexander, M., Campion, D., Cohen, D., Dikeos, D., Duan, J., Eichhammer, P., Godard, S., Hansen, M., Lerer, F.B., Liang, K.Y., Maier, W., Mallet, J., Nertney, D.A., Nestadt, G., Norton, N., O'Neill, F.A., Papadimitriou, G.N., Ribble, R., Sanders, A.R., Silverman, J.M., Walsh, D., Williams, N.M., Wormley, B., Arranz, M.J., Bakker, S., Bender, S., Bramon, E., Collier, D., Crespo-Facorro, B., Hall, J., Iyegbe, C., Jablensky, A., Kahn, R.S., Kalaydjieva, L., Lawrie, S., Lewis, C.M., Lin, K., Linszen, D.H., Mata, I., McIntosh, A., Murray, R.M., Ophoff, R.A., Powell, J., Rujescu, D., Van, O.J., Walshe, M., Weisbrod, M., Wiersma, D., Donnelly, P., Barroso, I., Blackwell, J.M., Bramon, E., Brown, M.A., Casas, J.P., Corvin, A.P., Deloukas, P., Duncanson, A., Jankowski, J., Markus, H.S., Mathew, C.G., Palmer, C.N., Plomin, R., Rautanen, A., Sawcer, S.J., Trembath, R.C., Viswanathan, A.C., Wood, N.W., Spencer, C.C., Band, G., Bellenguez, C., Freeman, C., Hellenthal, G., Giannoulatou, E., Pirinen, M., Pearson, R.D., Strange, A., Su, Z., Vukcevic, D., Donnelly, P., Langford, C., Hunt, S.E., Edkins, S., Gwilliam, R., Blackburn, H., Bumpstead, S.J., Dronov, S., Gillman, M., Gray, E., Hammond, N., Jayakumar, A., McCann, O.T., Liddle, J., Potter, S.C., Ravindrarajah, R., Ricketts, M., TashakkoriGhanbaria, A., Waller, M.J., Weston, P., Widaa, S., Whittaker, P., Barroso, I., Deloukas, P., Mathew, C.G., Blackwell, J.M., Brown, M.A., Corvin, A.P., McCarthy, M.I., Spencer, C.C., Bramon, E., Corvin, A.P., O'Donovan, M.C., Stefansson, K., Scolnick, E., Purcell, S., McCarroll, S.A., Sklar, P., Hultman, C.M., Sullivan, P.F., 2013. Genome-wide association analysis identifies 13 new risk loci for schizophrenia. Nat. Genet. 45, 1150–1159. Robins, E., Guze, S.B., 1970. Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am. J. Psychiatry 126, 983–987. Ruderfer, D.M., Fanous, A.H., Ripke, S., Mcquillin, A., Amdur, R.L., Gejman, P.V., O'Donovan, M.C., Andreassen, O.A., Djurovic, S., Hultman, C.M., Kelsoe, J.R., Jamain, S., Landen, M., Leboyer, M., Nimgaonkar, V., Nurnberger, J., Smoller, J.W., Craddock, N., Corvin, A., Sullivan, P.F., Holmans, P., Sklar, P., Kendler, K.S., 2014. Polygenic dissection of diagnosis and clinical dimensions of bipolar disorder and schizophrenia. Mol. Psychiatry 19 (9), 1017–1024. Sorensen, H.J., Mortensen, E.L., Reinisch, J.M., Mednick, S.A., 2004. Association between prenatal exposure to analgesics and risk of schizophrenia. Br. J. Psychiatry 185, 366–371. Sorensen, H.J., Mortensen, E.L., Reinisch, J.M., Mednick, S.A., 2009. Association between prenatal exposure to bacterial infection and risk of schizophrenia. Schizophr. Bull. 35, 631–637. Sorensen, H.J., Nielsen, P.R., Pedersen, C.B., Benros, M.E., Nordentoft, M., Mortensen, P.B., 2014. Population impact of familial and environmental risk factors for schizophrenia: a nationwide study. Schizophr. Res. 153 (1-3), 214–219. Szatkiewicz, J.P., O'Dushlaine, C., Chen, G., Chambert, K., Moran, J.L., Neale, B.M., Fromer, M., Ruderfer, D., Akterin, S., Bergen, S.E., Kahler, A., Magnusson, P.K., Kim, Y., Crowley, J.J., Rees, E., Kirov, G., O'Donovan, M.C., Owen, M.J., Walters, J., Scolnick, E., Sklar, P., Purcell, S., Hultman, C.M., McCarroll, S.A., Sullivan, P.F., 2014. Copy number variation in schizophrenia in Sweden. Mol. Psychiatry 19 (7), 762–773. Uggerby, P., Ostergaard, S.D., Roge, R., Correll, C.U., Nielsen, J., 2013. The validity of the schizophrenia diagnosis in the Danish Psychiatric Central Research Register is good. Dan. Med. J. 60, A4578. WHO, 1967. International Statistical Classification of Diseases, Injuries, and Causes of Death. Eight Revision (1965) Geneva. Wahlbeck, K., Forsen, T., Osmond, C., Barker, D.J., Eriksson, J.G., 2001. Association of schizophrenia with low maternal body mass index, small size at birth, and thinness during childhood. Arch. Gen. Psychiatry 58, 48–52.