Acta Ophthalmologica 2018

A nationwide population-based study of social

ABSTRACT. Introduction
Purpose: To study sociodemographic factors, associated diseases and survival of
Danish keratoconus patients. Keratoconus is the most common
Methods: All patients diagnosed with keratoconus 1977–2015 (n = 2679) were ectatic disease of the cornea. The dis-
matched to 10 persons who had not been diagnosed with keratoconus ease is usually bilateral though often
asymmetric, characterized by progres-
(n = 26 790). Conditional logistic regression assessed whether sociodemographic
sive steepening and thinning of the
factors and specific systemic diseases were associated with the odds of
cornea resulting in decreased visual
keratoconus. Mortality was assessed with time-to-event analysis. acuity. Keratoconus was previously char-
Results: After adjustment, non-Europeans had more than threefold higher odds acterized as a noninflammatory dis-
of keratoconus compared to Europeans (OR, 3.34; 96% CI 2.94–3.80). Single ease, but recent studies have indicated
persons had 27% higher odds (OR, 1.27; 95% CI 1.13–1.43), and divorced that an inflammatory component may
persons had 18% lower odds (OR 0.82; 95% CI 0.68–0.97) of keratoconus also be present (Fan Gaskin et al.
compared with persons in a relationship. Persons living in cities with <500 and 2015; Galvis et al. 2015). There fur-
500–4999 inhabitants had 40% (OR, 0.60; 95% CI 0.51–0.71) and 30% (OR, thermore seems to be a higher ratio of
0.70; 95% CI 0.61–0.81) lower odds of keratoconus, respectively, compared with males with keratoconus compared to
those living in the capital (>1 000 000 inhabitants). Persons receiving govern- females (Pearson et al. 2000; Wagner
ment substitution had 68% higher odds of keratoconus (OR, 1.68; 95% CI 1.30– et al. 2007; Woodward et al. 2016).
2.17) compared to self-employed. Keratoconus patients had more than twofold Many different diseases and envi-
higher odds of asthma (OR, 2.21; 95% CI 1.91–2.55), more than threefold ronmental factors, including but not
higher odds of allergic rhinitis (OR, 3.44; 95% CI 2.75–4.30), more than limited to eye rubbing, eczema and
atopy, Down syndrome, parental con-
sevenfold higher odds of atopic dermatitis (OR, 7.97; 95% CI, 6.21–10.21) and
sanguinity, mitral valve prolapse, sleep
69% higher odds of depression (OR, 1.69; 95% CI 1.18–2.43). Mortality rates
apnoea and connective tissue diseases,
were similar among keratoconus patients and controls (HR, 1.02; 95% CI 0.90– have been reported to be associated
1.16). with keratoconus (Gordon-Shaag et al.
Conclusion: Danish keratoconus patients differ from controls on several 2013; Patel & McGhee 2013). Most
sociodemographic factors and have higher risk of allergic rhinitis, asthma, studies are, however, characterized by
atopic dermatitis and depression. They do not have excess mortality compared to rather small populations of kerato-
controls. conus patients, typically from a single
cornea clinic. In 2016, Woodward et al.
Key words: epidemiology – keratoconus – mortality – register-based study (2016) published the first large-scale
register study based on health insur-
Acta Ophthalmol. ance billing records from 16 053 Amer-
ª 2018 Acta Ophthalmologica Scandinavica Foundation. Published by John Wiley & Sons Ltd ican keratoconus patients and their
matched controls. Their study showed
doi: 10.1111/aos.13961 that persons with asthma, collagen

1
Acta Ophthalmologica 2018

vascular disease, Down syndrome, to all residents at birth or immigration persons, matched on age and sex, was
sleep apnoea as well as Black and (Schmidt et al. 2014). This unique per- selected at random from the entire
Latino race compared to Caucasian sonal registration number was used to Danish population who had not been
had higher risk of keratoconus. They link data from the different registries. diagnosed with keratoconus at time of
further showed that patients with dia- The Danish Civil Registration System matching. Index time was defined as
betes mellitus as well as those living in contains information on date and place time of first keratoconus diagnosis for
large rural communities had lower risk of birth, gender, migration status, per- the keratoconus group and time of
of keratoconus. No significant associa- sonal registration number of parents as matching for the control group.
tion was found between keratoconus well as continuously updated informa-
and allergic rhinitis, mitral valve disor- tion on vital status, place of residence
Identification of possible associated
der, aneurysm, depression, education and civil status among others. The
diseases
level and level of income. Danish National Patient Register
Knowledge of the association of (NPR) is updated continuously and The possible associated diseases
both sociodemographic and socio-eco- contains information on all contacts to included in the analysis were identified
nomic factors as well as associated Danish hospitals including patients from a review of the literature as well
diseases is important for the further admitted to somatic wards (from as based on clinical observation. Breast
quest to understand the pathophysiol- 1977) as well as outpatient and emer- cancer was also included because pro-
ogy of keratoconus and to treat or gency visits (from 1995; Schmidt et al. lactin-induced protein, which is a
possibly even prevent the cause of the 2015). Contacts are coded with diag- marker for some types of breast cancer,
disease in the future. nostic codes classified according to the has been suggested as a possible
The life expectancy of keratoconus Danish version of the International biomarker for keratoconus (Priyadar-
patients is an area of controversy. Classification of Disease, 8th revision sini et al. 2014; Sharif et al. 2018).
Moodaley et al. (1992) showed that (ICD-8) 1977–1993 and 10th revision The NPR was used to identify individ-
292 keratoconus patients in their clinic (ICD-10) 1994 – present. Statistics uals diagnosed with the following dis-
had mortality rates similar to the gen- Denmark, a government institution eases: allergic rhinitis, asthma, atopic
eral population. In recent years, spec- providing statistics on the Danish pop- dermatitis, breast cancer, depression,
ulations of increased mortality among ulation, provided sociodemographic diabetes types 1 and 2, hypothyroidism,
keratoconus patients have arisen based information on each participant such inflammatory bowel disease, migraine,
on clinical observations of fewer kera- as level of income, source of income, mitral valve insufficiency, rheumatoid
toconus patient >50 years of age educational information and size of arthritis and sleep apnoea (ICD codes
(Pobelle-Frasson et al. 2004; Ertan & residence city. available in Table S2).
Muftuoglu 2008; Mcmonnies 2009; The possible associated diseases
McMonnies 2013). To our knowledge, were not restricted to occur either
Identification of keratoconus diagnoses
no time-to-event study comparing mor- before or after the keratoconus diag-
tality of keratoconus patients with a All patients registered with a kerato- nosis, as we were not estimating causal
matched control group has been con- conus diagnosis (ICD-8 code 378.79 associations between keratoconus and
ducted. It is vital to know whether and ICD-10 code H18.6) between 1977 the other diseases. We did, however,
keratoconus patients as a group have and 2015 were included in this study. perform a subanalysis restricted to
increased mortality or whether perhaps The first occurrence of either code in diseases diagnosed before and after
the lack of older keratoconus patients the NPR was used as the date of index time, respectively.
can be explained by other factors. Other diagnosis. Patients, who received the
explanatory factors could include diagnosis of Down syndrome (ICD-8
Statistical analysis
increased incidence due to more sensi- codes 310.50, 310.51, 311.50, 311.51,
tive diagnostic tools or simply decreased 312.50, 312.51, 313.50, 313.51, 314.50, Sociodemographic baseline data are
compliance with ophthalmology con- 314.51, 315.50, 315.51, 759.39 and presented as absolute values and per
trols due to a more stable disease later ICD-10 codes Q90, Q90.0, Q90.1, cent. Comparisons between the kera-
in life. Q90.1A, Q90.2, Q90.2A, D90.9, toconus groups and the control
The aim of this study was to char- Q90.9A and H19.8A) at any time group were performed by chi-squared
acterize the population of keratoconus point, were excluded from the main tests or Fisher’s exact test when
patients in Denmark in terms of analysis, because we considered this absolute values were below 20. Age
sociodemographic factors, associated group of patients to be fundamentally in years is presented as mean and
diseases and mortality. different from the remaining group of standard deviation (SD), and annual
keratoconus patients. However, sensi- income is categorized into quartiles.
tivity analyses were performed using all The distribution of socio-economic
Materials and Methods patients. Keratoconus patients who variables in the 10 years after index
had received corneal laser refractive time was investigated by percentage
Data source
surgery prior to keratoconus diagnosis of individuals at each category and
This nationwide population-based were likewise excluded to avoid includ- restricting to individuals with index
study included data from several Dan- ing patients with post-laser refractive time before 2005. Possible associated
ish national registries from 1977 to surgery ectasia. diseases are presented as absolute
2015. In Denmark, a unique civil per- For each patient with a diagnosis of values and percentage, and the two
sonal registration number is assigned keratoconus, a control group of 10 groups compared by means of

2
 Acta Ophthalmologica 2018

conditional logistic regression mod- Table 1. Sociodemographic data at year of KC diagnosis by case status.
els, with estimated odds ratios (OR)
Controls Keratoconus p-Value
and 95% confidence intervals (CI).
The overall OR of both sociodemo- Number (%) 26 790 2679
graphic factors and associated disease Gender, male (%) 17 910 (66.9) 1791 (66.9) 1.00
was adjusted for ethnicity, catego- Mean age at KC diagnosis, years (SD) 38.2 (15.9) 38.2 (15.9) 1.00
rized as European and non-Eur- Treatment, n (%)
opean. The choice of factors to Cornea transplantation – 646 (24.1%)
include in the adjusted analysis was Intracorneal ring segments (ICRS) 25 (0.9%)
Corneal cross-linking (CXL) 266 (9.9%)
based on directed acyclic graphs.
Familiar KC
Two subanalyses were performed: Parents identified – 2412 (90.0%)
first, restricted to diseases diagnosed KC patients with ≥1 parent with KC 30 (1.2%)
before index time; second, restricted Origin, n (%)
to persons who were at risk of being n identified 26 790 (100) 2679 (100) <0.001
diagnosed with that particular disease Danish 23 939 (89.4) 2103 (78.5)
at index time (i.e. excluding persons Emigrants 2440 (9.1) 468 (17.5)
diagnosed before index time). As Emigrant descendants 411 (1.5) 108 (4.0)
Origin of emigrants and descendants, n (%)
follow-up time in this analysis could
n identified 2835 (99.4) 574 (99.7) <0.001
differ between groups, we also per- Europe 1604 (56.6) 212 (36.9)
formed a time-to-event analysis, Africa 230 (8.1) 67 (11.7)
which showed similar results (data North/Middle/South America 118 (4.2) 29 (5.1)
not presented). Finally, mortality was Asia and Pacific Ocean (incl. Middle East) 883 (31.2) 266 (46.3)
compared among the two groups by Civil status, n (%)
means of Cox proportional hazard n identified 24 867 (92.8) 2496 (93.2) 0.049
models, with results presented as Relationship 9830 (39.5) 975 (39.1)
Divorced 1812 (7.3) 151 (6.1)
hazard ratios (HR) with CI, and
Single 12 624 (50.8) 1318 (52.8)
using pseudo-observations to esti- Longest living in a relationship 601 (2.4) 52 (2.1)
mate cumulative mortality propor- Residence city size, n (%)
tions at 5, 10, 15, 20, 25 and n identified 21 024 (78.5) 2118 (79.1) <0.001
30 years after index time. Capital city area 4959 (23.6) 630 (29.8)
Statistical analyses were performed >100 000 2153 (10.2) 234 (11.1)
using Stata 14 (Stata Corporation, 20 000–99 999 3746 (17.8) 416 (19.6)
College Station, TX, USA). 5000–19 999 2828 (13.5) 291 (13.7)
500–4999 3859 (18.4) 311 (14.7)
<500 3479 (16.6) 236 (11.1)
Ethics Level of income, n (%)
n identified 24 657 (92.0) 2479 (92.5) 0.416
The study was conducted based on Quartile 1 6161 (25.0) 639 (25.8)
registry data using encrypted identifi- Quartile 2 6160 (25.0) 633 (25.5)
cation numbers with no access to Quartile 3 6196 (25.1) 586 (23.6)
personal identification of participants. Quartile 4 6150 (24.9) 621 (25.1)
All data were stored on the secure Source of income, n (%)
n identified 23 261 (87.1) 2339 (87.3) <0.001
servers of Statistics Denmark. The
Self-employed 919 (4.0) 88 (3.8)
study was approved by the Danish Wage earner 13 162 (56.6) 1179 (50.4)
Data Protection Agency (project num- Unemployed 610 (2.6) 53 (2.3)
ber: 1-16-02-270-17). Student grants and sickness support 3386 (14.6) 362 (15.5)
Government substitution, long-term 1762 (7.6) 310 (13.2)
Pension 1907 (8.2) 175 (7.5)
Results Other incl. <15 years of age 1515 (6.5) 172 (7.3)
A total of 2679 patients had been Highest accomplished education, n (%)
n identified 24 595 (91.8) 2424 (90.4) <0.001
diagnosed with keratoconus in Den-
Primary and lower secondary school 8405 (34.2) 837 (34.6)
mark between 1977 and 2015. Of these, Upper secondary school 2520 (10.3) 258 (10.7)
1791 (66.9%) were men and mean age Vocational training 8135 (33.1) 653 (27.0)
at diagnosis was 38.2 (SD 15.9) years. Short-cycle higher education 996 (4.1) 114 (4.7)
Overall, 646 (24.1%) keratoconus Medium-cycle higher education 3011 (12.2) 318 (13.1)
patients had been treated with corneal Long-cycle higher education 1452 (5.9) 234 (9.7)
transplantation, 25 (0.9%) had PhD education 76 (0.3) 10 (0.4)
received intracorneal ring segments,
Emigrants: Persons born outside Denmark, none of the parents are both Danish citizens and born
and 266 (9.9%) had received corneal in Denmark. Descendants: Born in Denmark, none of the parents are both Danish citizen and
cross-linking. Furthermore, 30 (1.2%) born in Denmark. Relationship: both unmarried and married relationships are included. Capital
had at least one parent with kerato- city area >1 000 000.
conus (Table 1).

3
Acta Ophthalmologica 2018

Fig. 1. (A) Residence city size. Stacked diagram of the residence city size from index year and the following 10 years. Data limited to persons who
have the possibility to have 10 years of follow-up. Fewer keratoconus patients living in the smaller cities of <500 and 500–4999 inhabitants, this
difference remains stable over the 10-year period. (B) Source of income. Stacked diagram of the different sources of income from index year and the
following 10 years. Data limited to persons who have the possibility to have 10 years of follow-up. A larger portion of keratoconus patients compared
to controls are on long-term government substitution; this difference remains stable over the 10-year period. A natural tendency over the years of
more persons on pension and fewer receiving student grants are observed both in the keratoconus and in the control groups. (C) Level of income.
Stacked diagram of the four level of income quartiles from index year and the following 10 years. Data limited to persons who have the possibility to
have 10 years of follow-up. No differences are observed between the keratoconus and control group over the 10-year period.

Sociodemographic factors (adjusted OR, 0.60; 95% CI 0.51–0.71) persons with a long-cycle higher educa-
and 30% (adjusted OR, 0.70; 95% CI tion had 75% higher odds of kerato-
The baseline sociodemographic vari- 0.61–0.81) lower odds of keratoconus, conus (adjusted OR, 1.75; 95% CI 1.48–
ables are shown in Table 1 and adjusted compared with those living in the capital 2.07) and persons with vocational train-
OR in Table 3. Non-Europeans had (>1 000 000 inhabitants). These differ- ing had 13% lower odds of keratoconus
more than threefold higher odds of ences were similar even 10 years after (adjusted OR, 0.87; 95% CI 0.77–0.98).
keratoconus compared to Europeans index time (Fig. 1A). With regard to the There was no statistical significant asso-
(adjusted OR, 3.34; 96% CI 2.94–3.80). source of income, those on government ciation between level of income and
Persons who were single had 27% higher substitution had 68% higher risk of keratoconus (Table 3, Fig. 1C).
odds (adjusted OR, 1.27; 95% CI 1.13– keratoconus (adjusted OR, 1.68; 95%
1.43), and persons who were divorced CI 1.30–2.17) compared to self-
Associated diseases
had 18% lower odds (adjusted OR 0.82; employed. In the 10-year period after
95% CI 0.68–0.97) of keratoconus com- index time, the number of patients on Number of persons with each disease
pared with persons in a relationship. long-term government substitution and unadjusted odds ratio are shown in
Persons living in cities/towns with <500 remained the same (Fig. 1B). In highest Table 2. Patients with keratoconus had
and 500–4999 inhabitants had 40% achieved education at time of diagnosis, a more than twofold higher odds of

4
 Acta Ophthalmologica 2018

Table 2. Associated diseases, numbers by case status and unadjusted odds ratio with 95% be diagnosed with keratoconus com-
confidence interval. pared to Europeans. This is in accor-
dance with existing literature, where
Other diseases, n (%) Controls Keratoconus Overall OR (95% CI)
keratoconus prevalence is reported to
Allergic rhinitis 329 (1.2) 107 (4.0) 3.36 (2.69–4.20) be considerably lower in Europe than
Asthma 1211 (4.5) 247 (9.2) 2.15 (1.87–2.49) in countries in the Middle East and
Atopic dermatitis 164 (0.6) 114 (4.3) 7.32 (5.73–9.35) Asia (Ihalainen 1986; Pearson et al.
Breast cancer 311 (1.2) 26 (1.0) 0.83 (0.55–1.25) 2000; Nielsen et al. 2007; Millodot
Depression 206 (0.8) 35 (1.3) 1.71 (1.19–2.45) et al. 2011; Hashemi et al. 2014).
Diabetes 1061 (4.0) 114 (4.3) 1.08 (0.88–1.32)
Keratoconus patients are at least
Type 1 233 (0.9) 20 (0.8) 0.86 (0.54–1.36)
Type 2 752 (2.8) 85 (3.2) 1.14 (0.90–1.44) equally as likely to have completed
Hypothyroidism 301 (1.1) 41 (1.5) 1.38 (0.99–1.92) educations of substantial length requir-
Inflammatory bowel disease 388 (1.5) 43 (1.6) 1.11 (0.81–1.52) ing extensive reading and have a level
Migraine 363 (1.4) 61 (2.3) 1.70 (1.29–2.24) of income similar to nonkeratoconus
Mitral insufficiency 100 (0.4) 10 (0.4) 1.00 (0.52–1.92) subjects. This is encouraging as their
Rheumatoid arthritis 369 (1.4) 44 (1.6) 1.20 (0.88–1.65) reduced visual acuity could be sus-
Sleep apnoea 414 (1.6) 48 (1.8) 1.16 (0.86–1.58) pected to affect their ability to com-
95% CI = 95% confidence interval, OR = odds ratio. plete an education and earn a living.
Diabetes types 1 and 2 do not add to diabetes total as some patients were just diagnosed with However, we also found that a sub-
diabetes without specified type. stantial portion of keratoconus
patients receive long-term government
asthma (adjusted OR, 2.21; 95% CI separately (Table 4). The sex difference sunstitution, indicating that these
1.91–2.55), more than threefold higher was not statistically significant patients were unable to provide for
odds of allergic rhinitis (adjusted OR, (p = 0.08). When including patients themselves. The substantial size of the
3.44; 95% CI 2.75–4.30) and more than diagnosed with Down syndrome in the government substitution may explain
sevenfold higher odds of atopic dermati- survival analysis, keratoconus patients why an association between keratoco-
tis (adjusted OR, 7.97; 95% CI, 6.21– had 16% higher mortality than controls nus and level of income was not found.
10.21; Table 3). The odds of depression (HR 1.16; 95% CI 1.03–1.30; Table S3). Woodward et al. (2016) also included
were not significantly different between education and income in their registry-
keratoconus and controls before index based study but did not find an asso-
time (OR, 1.20; 95% CI 0.64–2.24). ciation with keratoconus. However,
However, after index time, persons with
Discussion they noted themselves that their data
keratoconus had 108% higher odds of The Danish population-based registers were based on insurance claims and
depression compared to controls [OR: enabled us to study a large population therefore may not hold true in patients
2.08 (95% CI: 1.33–3.25)], yielding an of keratoconus patients and, equally without health insurance.
overall adjusted OR of 1.69 (95% CI important, to compare them with a Persons living in smaller cities had
1.18–2.43; Table 3 and Table S1). Ker- large matched control group. This lower odds of keratoconus compared to
atoconus patients had 106% higher odds study identified several key findings. persons living in the capital. Woodward
of being diagnosed with migraine in the First, keratoconus patients differed et al. (2016) showed similar results.
period before index time as compared to from controls on several sociodemo- Possible explanations include that resi-
controls (OR 2.06; 95% CI 1.49–2.84). graphic factors: more were single as dents of smaller cities are less likely to
However, after index time, the diagnosis compared to being in a relationship, have access to corneal specialist for
of migraine was equally likely for kera- fewer lived in smaller cities as opposed diagnosis. Another possibility is that
toconus patients and controls (OR, 1.10; to the capital, more were on long-term environmental pollutants in larger cities
95% CI 0.64–1.88), yielding an overall government substitution as compared may be a factor in the development of
adjusted OR of 1.69 (95% CI 1.28–2.23; to being self-employed, fewer had keratoconus, possibly through
Table 3 and Table S1). vocational training, and more had a increased eye rubbing. Lastly, the higher
There was no statistically significant long-cycle higher education as com- odds of keratoconus among persons
association between keratoconus and pared to primary and lower secondary with a long-cycle higher education could
breast cancer, diabetes, hypothy- school as their highest accomplished influence keratoconus patients to live in
roidism, inflammatory bowel disease, education. Secondly, keratoconus was larger cities where higher education is
mitral insufficiency, rheumatoid arthri- significantly associated with allergic more easily available.
tis and sleep apnoea (Table 3). rhinitis, asthma, atopic dermatitis and In this study, we also found that
depression. Thirdly, the mortality of single persons have higher odds and
keratoconus patients was similar to persons who are divorced have lower
Survival analysis
other Danes, when excluding persons odds of keratoconus compared to those
Overall, no difference in survival was with Down syndrome. in a relationship. Our data does not
found between keratoconus patients explain why more keratoconus patients
and controls (HR 1.02; 95% CI 0.90– are single or fewer are divorced. The
Sociodemographic factors
1.16), including when looking at men process and choice of finding a partner
(HR 0.95; 95% CI 0.81–1.12) and In the adjusted analysis, non-Eur- is complex. More studies, possibly with
women (HR 1.14; 95% CI 0.94–1.39) opeans were threefold more likely to a qualitative approach, are needed to

5
Acta Ophthalmologica 2018

Table 3. Adjusted conditional logistic regression estimating associations of covariates with found twofold higher odds of asthma
keratoconus. and 1.6-fold higher odds of allergic
rhinitis among keratoconus patients,
Variable Adjusted OR (95% CI) p Value
but no significant association between
Ethnicity atopic dermatitis and keratoconus.
Europe Ref Their diagnosis of atopic dermatitis
Non-Europe 3.34 (2.94–3.80) <0.001 and allergic rhinitis was by self-report-
Civil status ing questionnaire, which may have
Relationship Ref limited their recordings to only the most
Divorced 0.82 (0.68–0.97) 0.029
severe cases. In the light of our findings,
Single 1.27 (1.13–1.43) <0.001
Longest living in relationship 0.79 (0.56–1.11) 0.179 we believe it is appropriate to inquire
City size about atopic disease in keratoconus
Capital area (>1 000 000) Ref patients, as they may benefit from diag-
>100 000 0.86 (0.74–1.01) 0.075 nosis and proper treatment. Further-
20–99 999 0.93 (0.81–1.06) 0.260 more, from an ophthalmologic view it
5–19 999 0.89 (0.76–1.03) 0.112 may be advantageous that the atopy is
500–4999 0.70 (0.61–0.81) <0.001 well treated to avoid undesirable beha-
<500 0.60 (0.51–0.71) <0.001
viour such as eye rubbing.
Level of income
Quartile 1 Ref Interestingly, patients with mental
Quartile 2 1.06 (0.92–1.22) 0.415 depression had higher odds of kerato-
Quartile 3 1.04 (0.90–1.21) 0.593 conus. Considering that this is hospital-
Quartile 4 1.18 (1.00–1.38) 0.049 diagnosed depression, these patients
Source of income will most likely be affected by depres-
Self-employed Ref sion in the more severe degrees. To the
Wage earner 0.97 (0.77–1.23) 0.825 best of our knowledge, no previous
Unemployed 0.87 (0.61–1.24) 0.443
study has shown an increased incidence
Student grants and sickness support 1.19 (0.90–1.56) 0.219
Government substitution, long-term 1.68 (1.30–2.17) 0.000 of depression among keratoconus
Pension 0.75 (0.51–1.12) 0.158 patients. Several studies have found
Other incl. <15 years of age 1.05 (0.77–1.44) 0.740 reduced quality of life among kerato-
Highest accomplished education conus patients, which may give a hint to
Primary and lower secondary school Ref their psychical well-being, but it is still a
Upper secondary school 1.00 (0.85–1.18) 0.997 way from diagnosed depression (Kymes
Vocational training 0.87 (0.77–0.98) 0.019 et al. 2004, 2008; Tatematsu-Ogawa
Short-cycle higher education 1.21 (0.97–1.50) 0.090
et al. 2008; Labiris et al. 2012; Saunier
Medium-cycle higher education 1.15 (0.99–1.33) 0.073
Long-cycle higher education 1.75 (1.48–2.07) <0.001 et al. 2017). Woodward et al. (2016) did
PhD education 1.28 (0.65–2.52) 0.482 not find an association between depres-
Allergic rhinitis 3.44 (2.75–4.30) <0.001 sion and keratoconus. However, their
Asthma 2.21 (1.91–2.55) <0.001 insurance claims data are prone to
Atopic dermatitis 7.97 (6.21–10.21) <0.001 selection bias in this regard. In the light
Breast cancer 0.81 (0.54–1.22) 0.314 of our findings, we believe it is appro-
Depression 1.69 (1.18–2.43) 0.004 priate to be aware of the psychical
Diabetes 1.03 (0.84–1.26) 0.761
health of keratoconus patients. Though
Type 1 0.87 (0.55–1.38) 0.558
Type 2 1.07 (0.85–1.36) 0.566 our data do not explain why kerato-
Hypothyroidism 1.30 (0.94–1.81) 0.115 conus patients are more likely to be
Inflammatory bowel disease 1.18 (0.86–1.62) 0.313 diagnosed with depression, we hypoth-
Migraine 1.69 (1.28–2.23) <0.001 esize that the impact of being diagnosed
Mitral insufficiency 0.97 (0.50–1.87) 0.923 with a chronic disease early in life, the
Rheumatoid arthritis 1.24 (0.90–1.71) 0.180 worry of reduced visual acuity and the
Sleep apnoea 1.18 (0.87–1.60) 0.298 burden of possible treatment such as
95% CI = 95% confidence interval, OR = odds ratio. corneal transplantation may be con-
tributing factors.
understand the background for these Several studies have shown associations Patients with migraine had higher
differences in civil status among kera- between keratoconus and these atopic odds of keratoconus. This result is
toconus patients. diseases, but to the best of our knowl- in line with Naderan et al. (2015), who
edge not to such large extends. Wood- showed that their 922 keratoconus
ward et al. (2016) showed 31% higher patients had higher risk of migraine than
Associated diseases
odds of keratoconus among asthma their controls. Migraine and kerato-
Patients with inflammatory atopic dis- patients, but found no association conus might be associated. However,
eases had more than threefold (allergic between keratoconus and allergic rhini- most of the association is explained by
rhinitis), more than twofold (asthma) tis. Merdler et al. (2015) based their migraine diagnosed before diagnosis of
and more than sevenfold (atopic der- study on the entry-level medical evalu- keratoconus. Therefore, we suspect that
matitis) higher odds of keratoconus. ation by the Israeli Defence Forces and the asthenopia associated with

6
 Acta Ophthalmologica 2018

Table 4. Mortality, hazard ratio and mortality per cent by case status. Strengths and limitations
Controls Keratoconus p-Value The main methodological strength of
this study is the use of the large,
Mortality, HR (95% CI) Ref 1.02 (0.90–1.16) longitudinal and very complete popu-
Males 0.95 (0.81–1.12) lation-based registries that are avail-
Females 1.14 (0.94–1.39)
able in Denmark. These registries,
Mortality, years after index date, % (95% CI)
5 years 5.16 (4.86–5.46) 4.88 (3.62–6.14) 0.564 combined with the free availability of
10 years 10.65 (10.17–11.13) 10.97 (8.87–13.07) 0.700 health care, minimize selection bias.
15 years 16.00 (15.29–16.71) 15.74 (12.94–18.55) 0.856 There are, nevertheless, also several
20 years 21.06 (20.22–21.91) 20.65 (17.04–24.26) 0.767 limitations. As with all registry-based
25 years 25.75 (24.65–26.84) 24.16 (19.71–28.61) 0.354 studies, we have to consider misclassifi-
30 years 30.80 (29.38–32.21) 28.55 (22.65–34.45) 0.330 cation. We were unable to confirm the
95% CI = 95% confidence interval, HR = hazard ratio.
diagnosis of keratoconus or indeed any
of the other associated diagnosis. How-
ever, with the NPR limited to cover
keratoconus is misdiagnosed as migraine frequently by an ophthalmologist,
hospital diagnoses, these patients have
before the correct diagnosis of kerato- which may increase their likelihood of
all been diagnosed in specialized settings,
conus is made. being diagnosed with keratoconus,
which we believe gives strength to the
We found no statistically significant thereby obscuring a true protective
correctness of the diagnosis. Further-
association between keratoconus and effect of diabetes.
more, because of the free availability of
diabetes. There have been conflicting Sleep apnoea and keratoconus were
health care, differential misclassification
reports about the association between not statistically significantly associated.
is unlikely.
diabetes and keratoconus. Seiler et al. Several studies, mainly from The United
The diagnosis of keratoconus is also
(2000) first reported 78% lower odds of States, have reported higher prevalence
given to patients who develop ectasia
keratoconus among persons with dia- of sleep apnoea among keratoconus
after laser refractive surgery, as there is
betes. However, their study was based patients. (Gupta et al. 2012; Saidel et al.
no separate ICD-10 diagnosis for this
on only two cases of diabetes in the 2012; Pihlblad & Schaefer 2013; Wood-
kind of ectasia. This prompted us to
keratoconus group and nine in the ward et al. 2016). In Denmark, the
review the charts of patients diagnosed
controls group. Woodward et al. awareness of sleep apnoea has been less
with ectasia post-laser refractive surgery
(2016) provided more power with more pronounced until recent years; there-
at the Department of Ophthalmology,
than 2000 diabetics in their keratoconus fore, we may have a larger proportion of
Aarhus University Hospital. Of 28
group; their data showed that uncom- undiagnosed patients. However, we do
ectasia patients, 24 had received the
plicated diabetic patients had 20% not suspect the diagnosis of sleep
diagnosis of keratoconus and, of these,
reduced odds and complicated diabetics apnoea to be dependent upon a previous
10 patients had also received a diagnosis
had 52% reduced odds of keratoconus. keratoconus diagnosis, why the impact
of laser surgery in our hospitals local
In contrast, Kosker et al. (2014) found of underdiagnosed sleep apnoea is
filesystem. This survey shows that we
that 6.7% of their 1377 keratoconus expected to be small.
have some contamination of post-laser
patients had type 2 diabetes as opposed
refractive surgery ectasia patients in our
to 4.8% of their controls (p = 0.005).
Survival analysis keratoconus group that we were not able
The pathophysiologic explanation put
to identify as they have not received the
forward to explain the lower number of To the best of our knowledge, this is the
diagnosis of laser refractive surgery,
diabetics with keratoconus is founded first study covering a whole population
because the laser refractive surgery is
on basic research showing that higher in a longitudinal design, including more
also performed in the private sector.
levels of blood glucose can promote than 2500 keratoconus patients com-
However, the number of ectasia patients
glycosylation of the collagen fibres of pared with a matched control group and
is very low and some of the ectasia
the cornea and thereby in effect stiffen with a follow-up period of up to
patients probably are keratoconus
the corneal tissue and prevent kerato- 36 years. Our findings of no excess
patients, who have been operated during
conus development (Dyer et al. 1993; mortality in the keratoconus group
the very early stages of their disease.
Spoerl et al. 1998). If higher blood (without patients with Down syndrome)
Therefore, we do not believe that this
glucose is a protecting factor for kera- are in line with the results of Moodaley
contamination of ectasia patients affects
toconus, then it may be hypothesized et al. (1992), who showed no excess
our data – and should it do so, we would
that Danish diabetic patients have bet- mortality of their 292 keratoconus
expect the ectasia patients to be similar
ter controlled blood glucose. With the patients. Although we expect the finding
to the healthy controls and thereby
Danish healthcare system solely funded to be transferable to other populations,
affect the data towards smaller differ-
by taxes and medication heavily substi- it should be considered that factors,
ences. It should further be noted that the
tuted, the financial burden of having such as social class and availability of
treatment of keratoconus patients is not
diabetes is very limited and therefore health care, influence life expectancy. If
handled in the private sector.
Danish patients may be more likely to keratoconus affects a person’s ability to
Another factor to consider is the
have better blood glucose control. keep a job, and health care is dependent
completeness of the data. As men-
Another explanation could be found upon employment status, then kerato-
tioned, the NPR only covers hospital
in surveillance bias. Because of their conus could indirectly have conse-
diagnosis, but the ophthalmological
diabetes, these patients are seen more quences for life expectancy.

7
Acta Ophthalmologica 2018

health care in Denmark is handled associated with higher risk of allergic Priyadarsini S, Hjortdal J, Sarker-nag A, Sejersen H, Asara
JM & Karamichos D (2014): Gross cystic disease fluid
both in the primary and in the sec- rhinitis, asthma, atopic dermatitis and protein-15/prolactin-inducible protein as a biomarker for
ondary sector. This along with the depression. Thirdly, persons diagnosed keratoconus disease. PLoS ONE, 9: 1–7.
Saidel MA, Paik JY, Garcia C, Russo P, Cao D & Bouchard C
mentioned data breaks in the NPR with keratoconus did not have excess (2012): Prevalence of sleep apnea syndrome and high-risk
created three periods in our data. The mortality compared to controls. characteristics among keratoconus patients. Cornea 31:
three periods which were not signifi- 600–603.
Saunier V, Mercier AE, Gaboriau T, Malet F, Colin J,
cantly different from each other were as Fourni
e P, Malecaze F & Touboul D (2017): Vision-related
follows: Period 1 (1977–1993), only References quality of life and dependency in French keratoconus
patients: impact study. J Cataract Refract Surg 43: 1582–
patients who were referred to the hos- 1590.
pital and received a corneal transplan- Dyer DG, Dunn JA, Thorpe SR, Bailie KE, Lyons TJ, Schmidt M, Pedersen L & Sørensen HT (2014): The Danish
McCance DR & Baynes JW (1993): Accumulation of Civil Registration System as a tool in epidemiology. Eur J
tation were registered in NPR; Period 2 Maillard reaction products in skin collagen in diabetes and Epidemiol 29: 541–549.
(1994–2010), all patients referred for aging. J Clin Invest 91: 2463–2469. Schmidt M, Schmidt SAJ, Sandegaard JL, Ehrenstein V,
Ertan A & Muftuoglu O (2008): Keratoconus clinical findings
treatment or second opinion were reg- Pedersen L & Sørensen HT (2015): The Danish National
according to different age and gender groups. Cornea 27: patient registry: a review of content, data quality, and
istered in NPR; and Period 3 (2011– 1109–1113. research potential. Clin Epidemiol 7: 449–490.
2015), with the availability of corneal Fan Gaskin JC, Loh IP, McGhee CNJ & Sherwin T (2015): An Seiler T, Huhle S, Spoerl E & Kunath H (2000): Manifest
immunohistochemical study of inflammatory cell changes diabetes and keratoconus: a retrospective case-control
cross-linking, we expect that all and matrix remodeling with and without acute hydrops in study. Graefe’s Arch Clin Exp Ophthalmol 238: 822–825.
patients diagnosed with or suspected keratoconus. Investig Ophthalmol Vis Sci 56: 5831–5837. Sharif R, Bak-Nielsen S, Hjortdal J & Karamichos D (2018):
Galvis V, Sherwin T, Tello A, Merayo J, Barrera R & Acera A
of keratoconus were referred to hospi- (2015): Keratoconus: an inflammatory disorder? Eye 29:
Pathogenesis of keratoconus: the intriguing therapeutic
potential of prolactin-inducible protein. Prog Retin Eye
tal as these handle the treatment and 843–859. Res [Epub ahead of print].
have the tomographic scanners avail- Gordon-Shaag A, Millodot M, Essa M, Garth J, Ghara M & Spoerl E, Huhle M & Seiler T (1998): Induction of cross-links
Shneor E (2013): Is consanguinity a risk factor for in corneal tissue. Exp Eye Res 66: 97–103.
able. To confirm our expectation that keratoconus? Optom Vis Sci 90: 448–454. Tatematsu-Ogawa Y, Yamada M, Kawashima M, Yamazaki
patients are being referred to hospital, Gupta PK, Stinnett SS & Carlson AN (2012): Prevalence of sleep Y, Bryce T & Tsubota K (2008): The disease burden of
apnea in patients with keratoconus. Cornea 31: 595–599.
we conducted a questionnaire survey Hashemi H, Khabazkhoob M, Yazdani N et al. (2014): The
keratoconus in patients’ lives: comparisons to a Japanese
normative sample. Eye Contact Lens 34: 13–16.
among the 32 primary-sector ophthal- prevalence of keratoconus in a young population in Wagner H, Barr JT & Zadnik K (2007): Collaborative
mologists in the Central Denmark Mashhad, Iran. Ophthalmic Physiol Opt 34: 519–527. Longitudinal Evaluation of Keratoconus (CLEK) Study:
Ihalainen A (1986): Clinical and epidemiological features of methods and findings to date. Contact Lens Anterior Eye
Region. Of the 27 ophthalmologists keratoconus. Genetic and external factors in the pathogen- 30: 223–232.
who responded, 81.5% would refer esis of the disease. Acta Ophthalmol 178(Suppl): 1–64. Woodward MA, Blachley TS & Stein JD (2016): The
Kosker M, Suri K, Hammersmith KM, Nassef AH, Nagra PK
patients they suspected of keratoconus & Rapuano CJ (2014): Another look at the association
association between sociodemographic factors, common
systemic diseases, and keratoconus an analysis of a
either right away or after a second between diabetes and keratoconus. Cornea 33: 1–6. nationwide heath care claims database. Ophthalmology
follow-up. Of patients with noticeable Kymes SM, Walline JJ, Zadnik K & Gordon MO (2004): 123: 457–465.e2.
Quality of life in keratoconus. Am J Ophthalmol 138: 527–
keratoconus, 96.3% would refer 535.
patients either right away or after a Kymes SM, Walline JJ, Zadnik K, Sterling J & Gordon MO
(2008): Changes in the quality-of-life of people with Received on July 24th, 2018.
second follow-up. Based on these keratoconus. Am J Ophthalmol 145: 611–617. Accepted on September 28th, 2018.
results, we believe the completeness of Labiris G, Giarmoukakis A, Sideroudi H, Gkika M, Fanari-
our data is good, especially for patients otis M & Kozobolis V (2012): Impact of keratoconus,
cross-linking and cross-linking combined with photorefrac- Correspondence:
with marked keratoconus. It should tive keratectomy on self-reported quality of life. Cornea 31: Sashia Bak-Nielsen, MD
also be noted that the average age at 734–739. Department of Ophthalmology
Mcmonnies CW (2009): Where are the older keratoconus
first diagnosis in the NPR is 38 years, Aarhus University Hospital
patients? Cornea 28: 836.
which is somewhat later than we would McMonnies CW (2013): Quo vadis older keratoconus Noerrebrogade 44
have expected. Possible explanations patients? Do they die at younger ages? Cornea 32: 496–502. 8000 Aarhus C, Denmark
Merdler I, Hassidim A, Sorkin N, Shapira S, Gronovich Y & Tel: +45 28572403
may include both a delay of the Korach Z (2015): Keratoconus and allergic diseases among
Fax: +Xxxx
primary care ophthalmologist and a Israeli adolescents between 2005 and 2013. Cornea 34: 525–
Email: [email protected]
529.
general delay in keratoconus diagnosis, Millodot M, Shneor E, Albou S, Atlani E & Gordon-Shaag A
which is well known to be difficult to (2011): Prevalence and associated factors of keratoconus in
The study was funded by Synoptik-Fonden, Fight
Jerusalem: a cross-sectional study. Ophthalmic Epidemiol for Sight Denmark, Aarhus University and Einar
diagnose in early or mild cases. One
18: 91–97. Willumsens Mindelegat.
might also expect to observe a decrease Moodaley LC, Woodward EG, Liu CS & Buckley RJ (1992):
in average age at first diagnosis follow- Life expectancy in keratoconus. Br J Ophthalmol 76: 590–
591.
ing the availability of corneal cross-
linking; this was however not the case
Naderan M, Shoar S, Rezagholizadeh F, Zolfaghari M &
Naderan M (2015): Characteristics and associations of Supporting Information
keratoconus patients. Contact Lens Anterior Eye 38: 199–
as there was no significant difference in 205. Additional Supporting Information
average age at diagnosis between the Nielsen K, Hjortdal J, Nohr EA & Ehlers N (2007): Incidence
may be found in the online version of
three data periods as mentioned above. and prevalence of keratoconus in Denmark. Acta Oph-
thalmol Scand 85: 890–892. this article:
Patel D & McGhee C (2013): Understanding keratoconus:
Table S1. ICD-codes for the included
Conclusion what have we learned from the New Zealand perspective?
Clin Exp Optom 96: 183–187.
diseases specified by ICD-8 and ICD-
Pearson AR, Soneji B, Sarvananthan N & Sandford-Smith JH
In conclusion, this study identified (2000): Does ethnic origin influence the incidence or 10.
several key findings. Firstly, Danish severity of keratoconus? Eye (Lond) 14(Pt 4): 625–628. Table S2. Adjusted odds ratios before
Pihlblad MS & Schaefer DP (2013): Eyelid laxity, obesity, and
keratoconus patients differ from con- obstructive sleep apnea in keratoconus. Cornea 32: 1232– and after index date and overall.
trols on several sociodemographic fac- 1236. Table S3. Mortality, Hazard ratio and
Pobelle-Frasson C, Velou S, Huslin V, Massicault B & Colin J
tors, including source of income, (2004): K
eratoc^ one : que deviennent les patients ^ ag
es?
mortality percent by case status,
highest accomplished education and [Keratoconus: what happens with older patients?] J Fr including Down Syndrome.
civil status. Secondly, keratoconus is Ophtalmol Sep 27: 779–782.