216

JOURNAL OF INSURANCE MEDICINECopyright � 2001 Journal of Insurance MedicineJ Insur Med 2001;33:216–220

ORIGINAL ARTICLE

Mortality in Various Types of OsteogenesisImperfectaRichard B. Singer, MD; Simon A. Ogston, PhD; Colin R. Paterson, DMMortality Article 826M1

Background.—Osteogenesis imperfecta (OI) is a group of closelyrelated inherited diseases characterized by abnormal bone fragility.The current clinical classification system delineates 6 types, one ofwhich (type II) is so severe that mortality is 100%, either intrauterineor perinatal. The types are differentiated by clinical groups, by se-verity, and by the presence or absence of other features such as bluesclerae or dentinogenesis imperfecta. There are no known previousstudies of mortality in OI.

Results.—From a registry created in association with the BrittleBone Society, 743 patients with OI in England and Wales were ob-served in the period 1980–1993. These were classified into 3 groups(type IA, type III, and types IB, IVA, and IVB combined). Averageannual mortality rates were determined in each group by sex andattained age. These rates were compared with 1981 rates in the pop-ulation of England and Wales, matched by sex and age. Results aregiven in terms of exposures, observed and expected deaths, and 2indices of excess mortality: mortality ratios and excess death ratesper 1000 person-years.

Conclusion.—In type IA, 51.5% of the OI cases overall, there wasno significant excess mortality (mortality ratio 108%, based on 15deaths). In type III, on the other hand, excess mortality was veryhigh in children and still significantly high at ages 15–34 years. Inthe combined group of types IB, IVA, and IVB, the mortality ratiowas 157% in patients aged 45 and up (not significant at the 95%confidence level), but higher ratios at younger ages were significant,even though based on a total of only 5 deaths.

Addresses: Blueberry Lane, C11,Falmouth, ME 04105 (Dr Singer);Department of Epidemiology andPublic Health, Ninewells Hospitaland Medical School, University ofDundee, Dundee DD1 9SY, Scotland(Dr Ogston); University of Dundee,Dundee DD1 4HN, Scotland (DrPaterson).

Correspondent: Richard B. Singer,MD, Consultant to AAIM.

Key words: Osteogenesis imperfec-ta, typing, comparative mortality.

Received: April 10, 2001

Accepted: May 10, 2001

BACKGROUND

Osteogenesis imperfecta (OI) is a largegroup of heritable disorders characterized byabnormal fragility of bone.1 It is sometimescalled ‘‘brittle bone disease.’’ It is the mostcommon cause of fractures in childhood, butpatients may also have abnormalities in othertissues, including the teeth, joints, sclerae,

blood vessels, and skin. Estimates of theprevalence of the disorder in western Europeand North America vary between 1 in 10,000and 1 in 20,000.

It is now thought that the great majority ofcases of OI result from mutations affectingthe formation or structure of type I collagen.While several hundred distinct mutationshave been identified, it is usual to classify

SINGER ET AL—MORTALITY IN OSTEOGENESIS IMPERFECTA

217

Table 1. Sillence Classification of Patients With Osteogenesis Imperfecta

Type Clinical Features Inheritance

I* Mild to moderate severity; little impairment ofgrowth; sclerae are blue or gray through life

Autosomal dominant; new mutations frequent

II Very severe disease causing stillbirth or early neona-tal death

New mutation or parental mosaicism

III Severe disease with antenatal fractures in many; pro-gressive deformity common; severe impairment ofgrowth; sclerae often blue or gray; dentinogenesisimperfecta common

New mutation, parental mosaicism, or (rarely) autoso-mal recessive

IV* Mild to moderate severity; some have growth impair-ment; sclerae normal in older children and adultsbut may be pale blue in early childhood

Autosomal dominant; new mutations frequent

* Subdivided into A (no overt dentinogenesis imperfecta) and B (dentinogenesis imperfecta present).

Table 2. Distribution and Other Data of Osteogenesis Imperfecta Patients in This Study

OI Type Male Female TotalPercentage

of TotalExposure

(Patient-Years)Number

of Deaths

IA 171 272 383 51.5 3370 15

IBIVAIVB

264437

514633

779070

10.412.19.4

719977714

457

Subtotal 107 130 237 31.9 2410 16

III 53 70 123 16.6 1160 26

Total 331 412 743 100.0 6940 57

cases on clinical grounds as proposed by Sill-ence.2 Table 1 summarizes this classification.While both types I and IV are usually de-scribed as mild to moderate, type IV patientstend to have more fractures; within type I,patients with type IB tend to have more frac-tures than those with type IA.3

Prior to the publications from Dundee4,5

there has been no information on mortality inOI. These were based on a registry of casesmainly from the United Kingdom and the Re-public of Ireland, ascertained largely throughan association with the Brittle Bone Society.The registry was initiated in 1980 and cur-rently consists of over 1450 patients. In thisarticle, we have drawn on the database usedin a study of life expectancy4 to determine forinsurance purposes the mortality in the dif-ferent types of OI.

SUBJECTS STUDIED AND FOLLOW-UP

Patients with OI were drawn from a data-base prepared for a study of life expectancyin OI.4 The observation period was January 1,1980, to January 1, 1993, but patients regis-tered prior to 1980 were given that date forentry into follow-up. Patients were excludedif they were type II, if the type could not bedetermined, if they were not resident in En-gland or Wales, or if the diagnosis was onlymade after death. After these exclusions,there were 743 patients in England and Walesremaining for study. Distribution by type andsex is given in Table 2 along with other data(type II is omitted). Over 50% of the patientswere in the least severe type IA. The othertypes shown had a prevalence ranging from9.4 to 16.6% (type III). After examination of

JOURNAL OF INSURANCE MEDICINE

218

Table 3. Comparative Mortality in Osteogenesis Imperfecta Type IA—383 Patients in England and Wales, 1980–93

AttainedAge or Sex

Exposure(Patient-Years)

E

ObservedDeaths

d

ExpectedDeaths*

d�

MortalityRatio %100d/d�

Mean Annual Mortality Rate/1000

Observedq

Expectedq�

Excess100(q � q�)

0–2425–4445 up

MaleFemale

Total

17281005638

148318873370

11

13

87

15

0.780.97

12.16

5.728.19

13.91

128103107

14085

108

0.51.0

20.0

5.43.74.5

0.61.0

19.0

3.94.34.1

�0.10.01.0

1.5�0.6

0.4

* Basis of expected deaths is 1981 English Life Tables.

Table 4. Comparative Mortality in Osteogenesis Imperfecta, Types IB, IVA, and IVB Combined—237 Patients inEngland and Wales, 1980–93

AttainedAge or Sex

Exposure(Patient-Years)

E

ObservedDeaths

d

ExpectedDeaths*

d�

MortalityRatio %100d/d�

Mean Annual Mortality Rates/1000)

Observedq

Expectedq�

Excess(q � q�)

0–2425–4445 up

MaleFemale

Total

1311705394

104113692410

23

11

97

16

0.600.736.99

3.055.268.31

335410157

295133193

1.54.3

28

8.65.26.6

0.51.0

18

2.93.93.4

1.03.3

10

5.71.33.2

* Basis of expected deaths is 1981 English Life Tables.

the mortality in these rather small groups ofpatients, we decided to show the higher mor-tality in type III separately but to combinepatients with OI types IB, IVA, and IVB intoa single group, with a prevalence of 39.9%and 16 of the total 57 deaths observed in anoverall exposure of 6940 patient-years. Intype IA, there were 15 deaths, and in type III,there were 26 deaths in a much smaller ex-posure. Females outnumbered males by aconsiderable margin. Mean attained age was24.0 years for the male and 26.8 for the femaleOI patients.

RESULTS

Excess mortality was minimal and not sta-tistically significant in the type IA patients,with 15 observed and 13.91 expected deaths

and an overall mortality ratio (MR) of only108% and excess death rate (EDR) of 0.4 extradeaths per 1000 per year (Table 3). There wasa sex difference, however, with an MR of140% for the male patients (not significanteven at a 10% significance level) and 85% forthe female patients. There were only 2 deathsunder age 45 so that the overall experience isbased predominantly on adults 45 years andolder.

As shown in Table 4, for the combinedgroup, types IB, IVA and IVB, excess mortal-ity was significantly high for males (MR295%) but not for females (MR 133%). Over-all, with 16 observed and 8.31 expecteddeaths, the MR was barely significant at the5% significance level (just outside 95% confi-dence limits). Despite only 5 deaths in the 2younger age groups, the excess mortality was

SINGER ET AL—MORTALITY IN OSTEOGENESIS IMPERFECTA

219

Table 5. Comparative Mortality in Osteogenesis Imperfecta, Type III—123 Patients in England and Wales,1980–93

AttainedAge or Sex

Exposure(Patient-Years)

E

ObservedDeaths

d

ExpectedDeaths*

d�

MortalityRatio %100d/d�

Mean Annual Mortality Rate/1000

Observedq

Expectedq�

Excess(q � q�)

0–45–9

10–3435 up

MaleFemale

Total

20821464889

551609

1160

11861

71926

0.4110.0480.2940.344

0.6180.4191.097

26,00018,600

2100290

113045002400

53379.3

11.6

12.73123

2.00.20.53.9

1.10.70.9

51378.87.7

11.63022

* Basis of expected deaths is 1981 English Life Tables.

high enough to be statistically significant, butthis was not true for the MR of 157% in pa-tients age 45 and up, based on 11 observeddeaths. The overall annual EDR was 5.7 per1000 for male patients but only 1.3 per 1000for females.

Almost all of the exposures and deathswere found at ages under 35 years in the typeIII cases (Table 5). Extremely high and signif-icant excess mortality was present in childrenunder age 15 years. In children 0–4 years, theMR was 26,000% and the annual EDR 51 per1000; for children age 5–9 years, these indiceswere 18,600% and 37 per 1000, respectively.The MR of 2100% at ages 10–34 years wasalso significant, despite its dependence ononly 6 deaths. At ages 35 years and up, onlya single death was observed. Despite the re-sults cited above for all other types of OI,EDR was higher in females (30 per 1000) thanin males (11.6 per 1000). The correspondingdifference in MR was exaggerated by the low-er mortality in females; MR was 4500% in fe-males and only 1130% in males, all ages com-bined.

In the single-page article published in theBritish Medical Journal,4 Paterson et al provid-ed results in terms of life expectancy in 4graphs, both male and female for type III andthe combined group (no graph was shown fortype IA because the authors ‘‘. . . could notdistinguish mortality in these patients fromthat in the general population.’’ Additional

unpublished data of Paterson et al. have beenutilized in this report. In the published arti-cle,4 values of life expectancy were shown atages 0, 1, 5, 10, 15, 25, 35, 45, and 55 yearsfor type III and for the additional age of 65in the combined group. The population lifeexpectancies were taken from the 1981 En-glish Life Tables. We have not shown the lifeexpectancy results in this report because lifeexpectancy is never used in risk classificationfor life insurance. The reader who might havea structured settlement annuity case in whichthe applicant has OI may consult the originalarticle.4

COMMENT

With the very limited numbers of deaths inmany of the subgroups, this study is an ex-cellent example of ‘‘the value of small class-es,’’ a topic discussed at an ALIMDA meetingprior to World War I by Dr Edwin C.Dwight,6 the Medical Director of the New En-gland Mutual Life Insurance Company, whowas an active member of the first Medico-Ac-tuarial Study Committee. Statistical signifi-cance, assessed by the Poisson distribution,has been shown to exist for some MR valuesin this article with as few as 6 deaths (typeIII patients age 15–34 years, Table 5).

Readers who wish to apply these results toconstruction of ratings for an underwritingmanual should be cautioned that MR values

JOURNAL OF INSURANCE MEDICINE

220

based on population-expected rates must beadjusted upward to allow for the fact that se-lect insurance mortality rates are much lowerthan population mortality rates. The adjust-ment process to convert population-expectedMRs to numerical ratings has been describedin detail.7 For example, in type IA OI patients,the sex difference in excess mortality shouldbe noted. With an MR of 85% for female pa-tients, it is highly probable that the overallMR based on select mortality would be with-in a standard limit of 125%. However, the MRfor male patients was 140%, with an EDR of1.5 per 1000. If we assume an average age of25 years and duration of 5–10 years, the ex-pected select rate from Table 5 of the citedarticle is 0.96 per 1000. The select MR wouldthen be (EDR � selectq�)/selectq� � (1.5 �1)/1 � 250%, not 140% (this approximate MRwould be even higher if the 1990–95 selectrate had been used). Male patients with typeIA OI therefore do not appear to be eligiblefor standard life insurance, as female patientswould be. Type III patients are generally de-

clinable for life insurance because of the veryhigh level of their excess mortality, but aftersuitable adjustment, many patients of typesIB, IVA, and IVB would probably be eligiblefor life insurance with an appropriate rating.

REFERENCES

1. Paterson CR. Osteogenesis imperfecta and otherheritable disorders of bone. Baillieres Clin EndocrinolMetab. 1997;11:193–213.

2. Sillence D. Osteogenesis imperfecta: an expandingpanorama of variants. Clin Orthop. 1981;159:11–25.

3. Paterson CR, McAllion S, Miller R. Heterogeneity ofosteogenesis imperfecta type I. J Med Genet. 1983;20:203–205.

4. Paterson CR, Ogston SA, Henry RM. Life expectan-cy in osteogenesis imperfecta. BMJ. 1996;312:351.

5. McAllion SJ, Paterson CR. Causes of death in osteo-genesis imperfecta. J Clin Path. 1996;49:627–630.

6. Dwight EC. The value of small classes. Proc AssocLife Insur Med Dir Am. 1913;15:210–218.

7. Singer RB. The conversion of mortality ratios to nu-merical ratings. In: Singer RB, Kita MW, and AveryJR, eds. Medical Risks—1991 Compend of Mortalityand Morbidity, Westport, Conn: Praeger Publishers;1994:66–73.