Parental history and venous thromboembolism: a nationwide study of age-specific and sex-specific familial risks in Sweden

Authors


Bengt Zöller, Center for Primary Health Care Research, CRC, Building 28, Floor 11, Entrance 72, Malmö University Hospital, S-205 02 Malmö, Sweden.
Tel.: +46 70 6691476; fax: +46 40 391370.
E-mail: bengt.zoller@med.lu.se

Abstract

Summary. Background: The value of parental history as a risk indicator for venous thromboembolism (VTE) has not been determined in a nationwide setting. Objectives: To perform the first nationwide study of age-specific and sex-specific familial VTE risks in offspring of parents hospitalized for VTE. Patients/Methods: The Swedish Multigeneration Register of 0–75-year-old subjects was linked to the Hospital Discharge Register for 1987–2007. Standardized incidence ratios (SIRs) were calculated for individuals whose parents were hospitalized for VTE as compared with those whose parents were unaffected. Results: Among 45 362 hospitalized offspring cases with VTE, 4865 offspring of affected parents were identified with a familial SIR of 2.00 (95% confidence interval [CI] 1.94–2.05). Familial SIR was slightly higher for male offspring than for female offspring (2.08, 95% CI 2.00–2.16 vs. 1.91, 95% CI 1.84–1.99). The risk in offspring was further increased when both parents were affected (3.97, 95% CI 3.40–4.61), with high familial risks at ages 20–29 years (10.00, 95% CI 5.91–15.84). The familial risks for VTE among offspring were increased from the age of 10 years up to 75 years, with familial SIRs of 3.96 (95% CI 3.13–4.94) at age 10–19 years and 1.48 (95% CI 1.17–1.84) at ages 70–75 years. However, the absolute incidence rate increased with age. Conclusions: Parental history is potentially useful for risk assessments of VTE, although age needs to be considered. Our results support the use of an age-dependent multicausal model to estimate the risk of VTE.

Introduction

Venous thromboembolism (VTE) affects approximately one per 1000 individuals per year [1–3]. The cause of VTE is multifactorial, with involvement of gene–gene and gene–environment interactions [4]. Risk factors for VTE include age, immobilization, surgery, trauma, pregnancy, puerperium, lupus anticoagulant, malignant disease, oral contraceptives, and postmenopausal hormone replacement [4]. Familial thrombophilia, for example clustering of VTE, was recognized at the beginning of the 19th century [5]. Heritability of VTE, that is, the proportion of the variance attributable to genetic effects on VTE, has been estimated to be 50–60% [6–8]. Five major genetic risk factors for VTE have been associated with familial thrombophilia: deficiencies of antithrombin, protein C, and protein S, and resistance to activated protein C caused by the factor V Leiden and prothrombin 20210A mutations [4,9]. The predictive value of family history for detecting any of these genetic risk factors is, however, low [10,11,13–15], suggesting that other genetic or non-genetic familial factors may be important. Several other weaker gene variants have been associated with VTE, although their contributions to familial thrombophilia have not been determined [16,17]. However, the traditional multicausal model of thrombosis does not explain why the same risk factors do not cause thrombosis in children but provoke this disorder in older people [4]. Rosendaal [4] has therefore suggested a dynamic age-dependent multicausal model that allows for interactions of various risk factors.

Few studies have investigated the association of familial history and risk of VTE [15,18]. In three case–control studies, family history increased the risk of VTE ∼ 2.5-fold, with odds ratios of 2.2, 2.3, and 2.7 [15,18]. However, no study has investigated the importance of a parental history of VTE as a risk indicator and the associated age-specific and sex-specific familial risks in offspring. Another novel contribution of this study is its approach; it was based on a nationwide register of all hospitalizations in Sweden between 1987 and 2007. The use of hospitalized cases eliminated any potential selection and recall bias. The Swedish family dataset, that is, the Multigeneration Register, is a validated source that has been proven to be reliable in the study of many familial diseases [20–23]. In this nationwide study, we have analyzed the age-specific and sex-specific familial risks of hospitalization for VTE among offspring of affected parents, which have not been determined before.

Materials and methods

This study was approved by the Ethics Committee of Lund University, Sweden. The VTE research database used for this study was constructed by linking several national Swedish registers, based on the MigMed2 datasets at the Center for Primary Health Care Research, Malmö, Lund University. Statistics Sweden, the Swedish government-owned statistics bureau, provided the Multi-Generation Register, in which offspring (second generation) born in Sweden in 1932 and later are linked to their parents (first generation), and are registered shortly after birth. Families could be defined by linking all of the children to their parents. The second generation represented the present study population. Linkages were made with the National Census Data in order to obtain individual socioeconomic status information. The final link was made by adding individual data from the Swedish Hospital Discharge Register, which records complete data on all discharges with dates of hospitalization and diagnoses since 1986. All linkages were performed by using an individual national identification number, which is assigned to each person in Sweden for their lifetime. This number was replaced by a serial number for each person, in order to provide anonymity. The serial number was used to check that each individual was only entered once for his or her first appearance with a VTE diagnosis. Over 11.8 million individuals in 3.9 million families were included in this database; 8.9 million individuals belonged to the second generation, in which the oldest born in 1932 had reached age 75 years at the end of the follow-up, which spanned from 1987 to 2007 [20–23]. The data in the MigMed Database are relatively complete.

Outcome variable

VTE patients were retrieved from hospital discharge data, reported according to the ninth (1987–1996) and 10th (1997–2007) versions of the International Classification of Diseases (ICD). In agreement with Souto et al. [6], VTE was defined as not only deep vein thrombosis (DVT) and pulmonary embolism (PE), but also superficial venous thrombosis and other forms of venous thrombosis. This definition of VTE is common in studies of familial thrombotic risk for defined genetic defects [24–26]. VTE was defined as the following ICD-9 and ICD-10 numbers: PE, ICD-9 415B and 416W, and ICD-10 I26; superficial or deep phlebitis or thrombophlebitis, ICD-9 451, and ICD-10 I80; portal vein thrombosis, ICD-9 452, and ICD-10 I81; other venous embolism or thrombosis, ICD-9 453, or ICD-10 I82; cerebral vein thrombosis and cerebral infarction resulting from cerebral vein thrombosis, ICD-9 437G, and ICD-10 I636 and I676; pregnancy-related VTE, ICD-9 671C, 671D, 671E, 671F, 671X and 673C, and ICD-10 O222, O223, O225, O228, O229, O870, O871, O873, O879, O882); and abortion-related VTE, ICD-9 639G, and ICD-10 O082 and O087). There is no official translation from ICD-9 to ICD-10. For this reason, DVT before delivery, code 671C in ICD-9, was defined by a grouping of ICD-10 codes (O223, O228, and O871), chosen to correspond as closely as possible to the previous code 671C in ICD-9. This is in agreement with the Swedish National Board of Health and Welfare ICD-9 to ICD-10 translator. A total of 45 362 patients were identified, on the basis of their first discharge recorded in the Hospital Discharge Register.

Individual variables included in the analysis

Sex was divided into males and females. Age at diagnosis was categorized in 10-year groups, and the groups were merged as necessary. Socioeconomic status for both males and females was divided into six groups according to occupation: (i) farmers; (ii) unskilled/skilled workers; (iii) white-collar workers; (iv) professionals; (v) self-employed; and (vi) all others [20–23]. Region was divided into three groups: (i) large cities, Stockholm, Gothenburg, and Malmo; (ii) southern Sweden; and (iii) northern Sweden. This allowed adjustment for regional differences in hospitalization [20–23]. Spouse was defined for the population over 25 years of age with same-parent children.

Statistical analysis

Person-years were calculated from the start of follow-up on 1 January 1987 until hospitalization for VTE, death, emigration, or the closing date, 31 December 2007. Age-specific incidence rates were calculated for the whole follow-up period, divided into 5-year periods. Standardized incidence ratios (SIRs) were calculated as the ratio of observed (O) to the expected (E) number of cases [27]. The expected number of cases was calculated for age (10-year groups), sex, period (5-year groups), region, and socioeconomic status-specific standard incidence rates. Familial risks in the offspring of affected parents were calculated for males and females, as compared with males or females whose parents were not affected by these conditions, using the cohort methods as previously described [28]. Ninety-five per cent confidence intervals (CIs) were calculated on the assumption of a Poisson distribution [28]. Age-adjusted incidence rates were calculated on the basis of the European standard population [29].

Results

We analyzed risks for offspring aged 0–75 years hospitalized for VTE in Sweden between 1987 and 2007. The follow-up period was 64 785 879 person-years for males without a parental history of VTE and 2 873 433 person-years for males with a parental history of VTE. The follow-up period was 61 815 359 person-years for females without a parental history of VTE and 27 47 968 person-years for females with a parental history of VTE. A total of 45 362 patients, 48.5% (n = 22 004) males and 51.5% females (n = 23 358), were diagnosed at a mean age of 50.7 years for males and 46.6 years for females. In total, 4865 (10.7%) of the 45 362 VTE cases among offspring occurred in those with a parental history of VTE. Of the male offspring cases, 2517 (11.4%) occurred among offspring with a parental history of VTE. Among females, 2348 (10.5%) cases occurred among offspring with a parental history of VTE.

Table 1 and Fig. 1 show the age-specific incidence rates for male and female offspring of parents with and without VTE. The overall incidence rate was higher for females than for males without a parental history (Table 1). No overall difference in incidence rate was observed between males and females with a parental history of VTE (Table 1). Both familial and non-familial incidence rates were higher for females than for males at the younger ages of 10–39 years. At age > 50 years, familial and non-familial incidence rates were higher for males than for females. The familial and non-familial incidence rates increased with age, and were highest among those aged 70–75 years (Table 1 and Fig. 1).

Table 1.   Age-specific number of venous thromboembolism (VTE) events and incidence rates (IRs) (per 100 000 person-years) in offspring
Age at diagnosis (years)MalesFemales
Without parental history of VTEWith parental history of VTEWithout parental history of VTEWith parental history of VTE
NIR95% CINIR95% CINIR95% CINIR95% CI
  1. CI, confidence interval.

< 10650.60.50.721.50.63.7490.50.30.600.0  
10–192832.62.32.92110.05.714.36546.45.96.85728.921.436.4
20–29110210.29.610.812231.726.137.4278227.426.428.428177.768.686.7
30–39206519.518.720.434154.448.760.2319231.830.732.941269.662.976.3
40–49412840.939.642.173595.288.3102.1402441.640.342.861282.876.289.3
50–59644780.978.982.9820145.6135.6155.6565272.870.974.6676123.7114.4133.0
60–694670145.1140.9149.2437247.1223.9270.33942121.0117.2124.8270152.7134.5170.9
70–75727231.4214.6248.239339.0232.6445.5715209.6194.2225.040322.0222.2421.8
All19 48730.129.730.5251787.684.291.021 01034.033.534.4234885.482.088.9
Figure 1.

 Age-specific incidence rates of venous thromboembolism in offspring by parental history.

Among children under 10 years of age, 111 had had VTE (Table 1). However, only two familial cases were observed for children < 10 years old (Table 2). There was a slightly lower risk of VTE among female offspring of mothers with VTE than for female offspring of fathers with VTE: SIR 2.03, 95% CI 1.91–2.16 vs. SIR 1.80, 95% CI 1.70–1.89 (Table 2). No difference was observed between male offspring of mothers with VTE and male offspring of fathers with VTE (Table 2).

Table 2.   Familial standardized incidence ratios (SIRs) for venous thromboembolism (VTE) in offspring of fathers/mothers with VTE
Age at diagnosis (years)Males having father with VTEMales havingmother with VTEFemales having father with VTEFemales having mother with VTE
OSIR95% CIOSIR95% CIOSIR95% CIOSIR95% CI
  1. O, observed number of cases; CI, confidence interval. Bold type: 95% CI does not include 1.00.

< 1013.220.0018.4612.330.0013.380   0   
10–19113.611.796.48113.881.936.97354.593.196.38223.151.974.77
20–29662.822.183.59633.142.414.021723.012.583.501202.442.032.92
30–391852.692.323.111652.432.072.832262.231.952.542082.171.892.49
40–493362.242.012.494322.272.062.502791.941.722.183541.921.722.13
50–593201.871.672.085281.861.702.022611.771.562.004351.741.581.91
60–691441.711.442.013061.741.551.95971.451.181.771791.241.071.44
70–7530.390.071.14361.831.282.53111.510.752.71291.470.992.12
All10662.091.972.2215422.031.932.1310812.031.912.1613471.801.701.89

Familial SIR was highest among offspring aged 10–19 years (3.96), and decreased with age to 1.48 at age 70–75 years (Table 3). Similar increased and overlapping age-dependent familial SIRs were observed in all age intervals from 10 to 75 years for both male and female offspring, although the overall familial SIR was slightly higher for males than for females with one affected parent: SIR 2.08, 95% CI 2.00–2.16 vs. SIR 1.91, 95% CI 1.84–1.99 (not shown in any table). Although the familial SIRs decreased with age (Tables 2 and 3), the familial incidence rate increased with age and was highest among those aged 70–75 years (Table 1 and Fig. 1).

Table 3.   Familial standardized incidence ratios (SIRs) for venous thromboembolism (VTE) in offspring of parents with VTE
Age at diagnosis (years)Offspring of parents with VTEOffspring with both parents having VTEMales with both parents having VTEFemales with both parents having VTE
OSIR95% CIOSIR95% CIOSIR95% CIOSIR95% CI
  1. O, observed number of cases; CI, confidence interval. Bold type: 95% CI does not include 1.00.

< 1021.570.155.790   0   0   
10–19783.963.134.9416.610.0037.91122.080.01126.540   
20–294032.822.553.111810.005.9115.84712.955.1326.84118.734.3315.68
30–397532.392.222.57315.513.747.8393.831.737.29226.734.2110.20
40–4913472.132.022.25544.203.165.49335.113.517.18213.292.035.03
50–5914961.831.741.93483.192.354.23283.552.365.14202.791.704.32
60–697071.541.431.66192.721.634.25133.471.845.9661.850.674.05
70–75791.481.171.840   0   0   
All48652.001.942.051713.973.404.61914.283.445.25803.672.914.57

Familial SIRs were the highest among all offspring with both parents affected (Table 3), especially at young ages. This same age-dependent pattern was observed for both female and male offspring with both parents affected (Table 3). The familial SIRs tended to be higher for male offspring than for female offspring with both parents affected; however, the 95% CIs overlapped (Table 3).

We tested for heterogeneity of the familial risks for the different manifestations of VTE. These risks were, however, similar for PE (1.94, 95% CI 1.85–2.04), deep or superficial thrombosis of the lower extremities (2.04, 95% CI 1.96–2.13), and other forms of venous thrombosis (1.99, 95% CI 1.87–2.12). Moreover, the familial risks for superficial thrombophlebitis (1.96, 95% CI 1.71–2.22) and DVT (2.07, 95% CI 1.97–2.18), were similar. Only 234 (4.8%) of the 4865 VTE offspring cases with a parental history of VTE had had superficial thrombosis, as compared with 2175 (4.8%) of all of the 45 362 offspring cases with VTE.

In order to test for the extent of environmental sharing of the observed risks, we calculated the risk of spouses being hospitalized for VTE. There was a small increased risk for VTE among spouses. The SIR was 1.08 (95% CI 1.03–1.13) for husbands and 1.06 (95% CI 1.01–1.11) for wives.

Discussion

The effect of age and sex on familial SIRs for offspring of parents with VTE is a novel finding. The present study demonstrates an age-dependent and sex-dependent influence on the familial thrombotic risk, and supports the use of a dynamic age-dependent multicausal thrombosis model [4]. The traditional multicausal model does not take age into consideration [4]. This large nationwide study of offspring of parents with a history of hospitalization for VTE in Sweden highlights the value of parental history as a predictor of the risk of VTE. Parental history of VTE appears to be an important risk factor for VTE among offspring, with a higher relative risk than for many other identified genetic risk factors [16,17]. Our investigation is the first nationwide study to demonstrate an increased familial risk of VTE among offspring aged 10–75 years (but not < 10 years) of parents with a history of VTE. Thus, parental family history of VTE is a potential clinically useful risk indicator for VTE, at least from ages 10 to 75 years. An interesting finding is that the familial SIRs decrease with age, but the absolute incidence rates increase. Thus, familial factors are also important at older ages. The occurrence of only two familial offspring cases among the children < 10 years old could be attributable to the small number of cases. It could also reflect the fact that the risk is very low in childhood, even among familial cases, and that, in most cases, one or more acquired risk factors are necessary for the development of VTE [30]. Our estimates of VTE in children are in agreement with a report from Canada, which found the incidence of DVT to be 0.7 per 100 000 person-years (Table 1) [31]. Our study shows the importance of choosing cases between ages 10 and 50 years with the highest familial SIRs in genome-wide association studies of VTE, in order to maximize the likelihood of identifying genetic defects causing familial thrombophilia [32].

Our study included all manifestations of VTE, and not only of DVT and PE. This contrasts with previous studies of familial histories of VTE, in which only PE and/or DVT have been studied [15,18]. This definition of VTE could be a potential limitation if the genetic background is different for different manifestations. We tested for heterogeneity of familial risks, but no significant differences were observed for the different manifestations of VTE. Only 4.8% had superficial thrombosis, and the risk was similar to those for other VTE manifestations.

There was a slightly increased relative risk for female offspring with a father with VTE as compared with those with a mother with VTE (Table 2). Speculatively, this could be the result of inheritance of some minor X-linked risk factor, sex-specific imprinting, or sex-specific familial environmental factors. However, the difference is small and is unlikely to be clinically important. A slightly higher overall risk was observed for familial SIRs for male offspring than for those of female offspring (SIR 2.08, 95% CI 2.00–2.16 vs. SIR 1.91, 95% CI 1.84–1.99). However, this sex-specific difference is also small and unlikely to be of clinical importance.

The low spouse effects of only 6% and 8% for wives and husbands, respectively, demonstrate that most of the familial risk of VTE in offspring is genetic and not an acquired familial effect. The especially increased risks among offspring having both parents affected by VTE suggest a gene–gene interaction or an additive genetic effect rather than a family environmental effect (Table 3). Moreover, at younger ages, these gene–gene effects appeared to act synergistically, with a multiplicatively increased risk among offspring with both parents affected.

The present design, focusing on hospitalized cases of VTE, has numerous advantages. One important advantage of our study is that it is nationwide and based on all hospitalized patients, without the risk of selection bias. The Swedish family dataset, that is, the Multigeneration Register, is a validated source that has been proven to be reliable in the study of many familial diseases [20–23]. In most existing studies, information on family cases of VTE is based on self-reported or proxy-reported data [7,15,18]. Such data may be subject to selection and recall bias, making quantitative estimates imprecise. The data in the MigMed Database are relatively complete. In 2001, personal code numbers were missing for only 0.4% and the main diagnosis for 0.9% of hospitalizations [22]. Information on occupational status, retrieved from the national census records in the database, was 99.2% complete [22]. Moreover, cases with cancers or other diagnoses were not excluded in the present study, in agreement with Bezemer et al. [15]. Thus, our data reflect the total impact of the familial parental history of VTE in the whole Swedish population up to age 75 years.

The advantages include complete nationwide coverage in a country with high standards in the medical diagnosis of parents and offspring made by a specialist during extensive examinations in the hospital. The hospitalization discharge register contains no information about diagnostic procedures. However, the overall diagnostic validity of the Swedish Inpatient Register is close to 90% [33]. Moreover, the diagnosis of VTE in the Swedish Inpatient Register has been validated and found to be correct in 95% of VTE diagnoses [34]. Moreover, in 91%, VTE is diagnosed by an objective method [34]. This is in agreement with three recent Swedish studies demonstrating that VTE is seldom diagnosed without objective investigations [35–37]. The high validity of the Swedish Inpatient Register for VTE is in conformity with studies that have found a validity of ∼ 95% for stroke and myocardial infarction [38–40]. A good concordance of 89% was found between hospital discharge diagnoses and the underlying causes of death of those who were hospitalized and later died under dramatic conditions [41]. As it is possible that the diagnostic accuracy could have varied between geographic regions, we adjusted for geographic region in order to minimize this possible bias. Only a main diagnosis of VTE was used in the present study, to further increase the accuracy. Moreover, our estimates of the familial and non-familial incidence rates at different ages are in agreement with reported overall incidence rates from other studies, and confirm the exponentially increased incidence rate with age [1–3,42], further supporting our model for studying the risk of familial VTE. Higher incidence rates have been reported by Heit et al. [43], especially for older men. However, there are differences in incidence rates for VTE between all reported studies [1–3,42,43]. This may be because of differences in the inclusion of not only hospitalized cases but also outpatients and autopsied cases. Moreover, the definition and method of diagnosis of VTE, age, and environmental and ethnic differences may explain the observed differences in incidence rates between studies [1–3,42,43]. However, this is a non-differential bias with regard to estimating familial risks in the present study.

There are some possible disadvantages of the present study. However, many of them involve a non-differential bias. The Swedish Hospital Discharge Register has complete data only from 1987, covering a period of 20 years. Thus, many VTE cases were unknown because the parents or offspring were hospitalized before the start of the follow-up in 1987. This could lead to an underestimation or overestimation of the familial SIR if cases were not lost in proportion to the estimated familial SIRs. This potential bias seems less likely. Our total estimate of a familial SIR of 2.0 is similar to reported odds ratios of 2.2, 2.3 and 2.7 for the risk of VTE attributable to a family history of VTE [15,18]. Thus, it seems that cases were lost in proportion to the estimated familial SIRs, and that this is a non-differential bias. Moreover, the risk of recurrent VTE is high [4,42]. Therefore, many undiagnosed cases will eventually be affected by a new thrombotic event during the follow-up period between 1987 and 2007, and classified as thrombotic cases. Our concept of studying familial risks has been used for many other complex diseases with correct familial risk estimates [20–23], and it is unlikely that this method yields incorrect estimates only for VTE and not other complex diseases. Moreover, published studies estimating familial risks based on a history of VTE among relatives are likely to suffer from recall bias regarding events many years earlier, and our approach is not biased by this problem. Another non-differential bias is the introduction of low molecular weight heparin (LMWH). Since the introduction of LMWH, an increasing number of DVT cases have been treated as outpatients. However, age-specific incidence rates were calculated for the whole follow-up period, divided into five 5-year periods, and an adjustment was made for possible changes in the incidence rate over time. A number of less severe thrombotic cases are treated as outpatient cases. If familial cases tend to be more severe than non-familial ones, this could lead to an overestimation of the VTE risk, but there is no evidence that strengthens this hypothesis. In any case, our data will still reflect the risk for severe hospitalized cases. Another potential disadvantage includes the possibility that some selective factors could operate in the process of hospitalization that would favor some families being hospitalized. The low spouse correlation failed to support any strong selection for hospitalization of certain families. Unfortunately, our data did not include environmental risk factors. As a compromise, we adjusted for socioeconomic status (occupation) in the models. We therefore cannot compare the importance of parental history of VTE in unprovoked and provoked cases. However, according to a recent publication, family history is important in both provoked and unprovoked cases [15]. This is in agreement with family studies of defined genetic defects, where approximately 50% of first thrombotic event cases are unprovoked [24–26,44]. Inclusion of only unprovoked cases may therefore give incorrect estimations of the overall familial risks. Thus, acquired risk factors are usually not taken into account when considering the overall importance of a family history of VTE or heritability of VTE [6–8,15,18], although familial risks have been estimated in unprovoked VTE cases by Couturaud et al. [45]. In that study, VTE was observed in a first-degree relative in 5.3% of thrombotic cases [45], which is much lower than in the present study, where 10.7% of thrombotic cases were familial, especially as our study only considered hospitalized histories of parental VTE. This difference may be attributable to recall or inclusion bias in the study by Couturaud et al. [4], which illustrates the benefits of our study design.

In conclusion, familial parental factors are important for the development of VTE among offspring, at least from the ages of 10 to 75 years. The relative contribution of familial parental factors decreases with age, although the absolute risk increases. In clinical practice, parental history is useful for risk assessment. Moreover, our results support the use of a time-dependent multicausal model to estimate the risk of VTE. The study shows the importance of choosing cases between the ages of 10 and 50 years in genome-wide association studies, in order to maximize the probability of identifying genetic defects causing familial thrombophilia.

Acknowledgements

The registers used in the present study are maintained at Statistics Sweden and the National Board of Health and Welfare. This work was supported by grants to K. Sundquist and J. Sundquist from the Swedish Research Council (2008-3110 and 2008-2638), the Swedish Council for Working Life and Social Research (2006-0386, 2007-1754 and 2007-1962), and the Swedish Research Council Formas (2006-4255-6596-99 and 2007-1352).

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Ancillary