Is Early Blindness Protective of Psychosis or Are We Turning a Blind Eye to the Lack of Statistical Power?

The means required for coping with schizophrenic patients who are also blind would tend to make them memorable. Yet men who had treated more than ten thousand patients had difficulty recalling one who was blind.

– Chevigny and Braverman (1950)

The words above suggest that blindness may be protective of schizophrenia, and represent the first formulation of this hypothesis, published in the book “The Adjustment of the Blind” from 1950 by Hector Chevigny and Sydell Braverman.¹ Since then, several authors have supported the hypothesis, based on the continuous lack of case stories, and have refined it, proposing that the protective effect may be restricted to congenital or early blindness (in contrast to blindness acquired later in life) and maybe even to congenital/early blindness caused by lesions in the visual cortex (in contrast to blindness arising from peripheral lesions in the eyeball).² Recently, the first large epidemiological study was conducted to test the hypothesis that blindness is protective of schizophrenia, by applying Western Australian whole-population register data.³ In a cohort of 467 945 subjects aged between 14 and 35 years at the time of psychiatric data extraction, none out of 66 children with cortical blindness, before age 6, had developed schizophrenia, nor any other psychotic illness, and 8 out of 613 children with congenital/early “peripheral blindness” (somewhat loosely defined as this included ophthalmological diagnoses, such as congenital glaucoma, lens malformations, and retinal dystrophy—conditions with varying degrees of visual impairment and, only sometimes, complete blindness) had developed a psychotic illness other than schizophrenia. The authors admitted to the study being possibly underpowered, but stated that their results do support the hypothesis that congenital/early cortical blindness is protective of schizophrenia. If true, this finding could potentially aid our understanding of the etiology of schizophrenia and has already sparked new theories—such as those elegantly put forward by Pollak and Corlett in this issue of Schizophrenia Bulletin⁴—and a great deal of interest from popular media.⁵ However, the observed lack of case reports and register-based concurrences of the 2 disorders could be due to pure chance, low power, and ascertainment bias. Indeed, demonstrating a negative association between two rare conditions is extremely challenging and requires large sample sizes—a challenge which may be met by the Danish national health registers.⁶

Accordingly, we performed a nationwide population-based study based on data from the Danish National Patient Register (containing diagnostic information regarding blindness) and the Danish Psychiatric Central Research Register (containing diagnostic information regarding mental disorders)⁶ to examine whether early blindness is protective of development of psychotic disorders. First, we identified all individuals born in Denmark between 1977 and 2016. Within this birth cohort, we identified cases of blindness (ICD-8: 379-, ICD-10: H54.0) diagnosed before the age of 6 years and cases of schizophrenia (ICD-8: 295, excluding 295.79, ICD-10: F20) or psychotic disorder more broadly defined (ICD-8: 295.x9, 296.89, 297.x9, 298.29–298.99, 299.04, 299.05, 299.09, 301.83, ICD-10: F2x) diagnosed before the end of follow-up (death, emigration, or December 31 2016, whichever came first). In the birth cohort consisting of 2 500 332 individuals followed for a total of 47.5 million person-years, we identified 460 cases of early blindness, 10 440 cases of schizophrenia, and 19 716 cases of psychotic disorder. Based on the 8100 person-years of follow-up for the individuals with early blindness (and the incidence rates of the mental disorders reported above), we would expect to find 1–2 cases of schizophrenia and 3–4 cases of psychotic disorder if there was no protective effect of early blindness. We found <5 with schizophrenia and <5 with any psychotic illness (numbers <5 are not publishable according to Danish legislation—due to the risk of identification of individuals). This means that we can neither confirm nor refute the hypothesis that early blindness is protective of psychosis, due to insufficient power. To achieve sufficient power, we would need an even larger cohort and/or longer follow-up time—but how much larger/longer? We estimated the required sample size, depending on how protective early blindness could be. Specifically, based on nationwide Danish register data from Pedersen et al,⁷ we assumed a lifetime cumulative incidence rate for schizophrenia of ∼2% (figure 1, green line), and ∼4% for psychotic disorders (figure 1, blue line). If early blindness (estimated cumulative incidence = 460/2.500.332 = 0.000184) was to completely abolish the risk of schizophrenia (hazard ratio ∼.00), significant results would require a cohort of approximately 3 000 000 individuals long enough to reach the assumed cumulative incidence, and if early blindness would only halve the risk (hazard ratio ∼.5), a cohort of at least 11 000 000 individuals would be required. Substantially larger cohorts would be required to investigate a protective effect of early cortical blindness, specifically, as this is a much more rare condition than early blindness in general. Such numbers could potentially be achieved by combining data from several nationwide registers with long follow-up.

Fig. 1.

Required sample size to detect statistically significant protective of early blindness against schizophrenia or psychosis. The required sample size depends on how protective early blindness is against schizophrenia or psychosis, represented by the hazard ratio on the x-axis. Accordingly, if the effect is weaker, a larger sample size is needed, and vice versa. The required sample size also depends on the background cumulative incidence of the outcome, indicated by the different colors, where the incidence of schizophrenia roughly corresponds to the green curve, and psychotic disorders to the blue curve.⁷ Statistics: log-rank test, Freedman method. Power = 0.9, alpha = 0.05, N2/N1 = 460/2.500.332 = 0.000184.

In conclusion, the results of this study, which is by far the largest of its kind with its follow-up time of 47.5 million person-years, show that the hypothesis of early blindness being protective of psychosis is currently untestable using Danish register data. As all other published results are also inconclusive,^1–3 we strongly advice against drawing any conclusions on the issue based on the available evidence. Furthermore, theories on how blindness is protective of psychosis⁴ are premature and should, if proposed, clearly state that there is currently no evidence that it actually is protective.

Funding

None.