Two hundred nine cases and 425 controls completed the survey. Participation rates were approximately 73% in cases and 54% in controls. All but 13 participants completed the survey by mail, the rest by telephone. We excluded a total of nine potential cases and 10 controls because they did not meet inclusion criteria. Most omitted cases were diagnosed with retinal holes rather than RD/T and the majority of omitted controls reported a past history of RD/T.

We collected complete data on 200 cases of RD (n = 105) or RT (n = 95). About half of these (n = 113) were drawn from the central Massachusetts health care system, and the rest from the independent retinology practices in the same catchment area. We also gathered data on 415 controls, from patients visiting ophthalmology/optometry clinics for routine eye examinations (Table 1).

Participants were predominantly white (controls 87%, cases 93%), which is consistent with the central Massachusetts population. Cases were somewhat more likely to be male (55% of cases vs 36% of controls). As expected, cases were older on average than controls (61 vs 54 years, respectively). We observed a nonlinear shape of the relationship between age and RD/T (Fig. 1). The risk of RD/T rose steadily from young adulthood and peaked at about age 65, after which it declined rapidly through at least age 80. We investigated several ways to represent this nonlinear relationship in our multiple logistic regression models, including a three-category class variable, a spline, and a quadratic form using linear and squared terms. These three forms all fit similarly and here we report results from the quadratic parameterization.

Among myopics, we found almost a three-fold increase in risk of RD and only a weak increase in risk of RT compared to non-myopics (Table 2). Subanalyses restricted to participants for whom we had refractive error data to define myopia showed a very similar association with risk of RD/T. Also, controlling for confounding by myopia measured with refractive error rather than self-report had similar effects on other risk factors, suggesting that the self-reported measure was not introducing bias into the multivariable models.

Previous cataract surgery (in the same eye) greater than three months before RD/T was a strong risk factor for both RD and RT, although the magnitude of the effect was somewhat stronger for RD than for RT (OR = 10.4, 95% CI 4.8 to 22.1) and (OR = 6.5, 95% CI 2.7 to 15.4), respectively (Table 2). Family history of a close relative with RD/T was associated with more than a two-fold increase in risk of RD/T.

Increasing BMI was associated with lower risk of both RD (OR = 0.94, 95% CI 0.89 to 0.98) and RT (OR = 0.93, 95% CI 0.89 to 0.98). This negative association was unanticipated because the two previous studies reported positive associations.^5,7 We investigated the hypothesis that the negative BMI – RD/T association might be a secondary effect of a disproportionately high prevalence of diabetics among the controls (in this insured population, diabetics may be more likely to have regular eye examinations, and tend to be overweight). We therefore restricted the analysis to nondiabetics, and found that the negative BMI – RD/T association was of similar magnitude, but with wider CI (OR = 0.96, 95% CI 0.92 to 1.00), suggesting that the negative BMI – RD/T association was not fully explained by diabetes.

Although the diagnosis and treatment of RD/T are highly developed and generally effective, there has been relatively little research on primary prevention. We identified seven different risk factors for RD/T, with various implications for prevention.

Age is a well-known and strong risk factor for both RD and RT; however, the nonlinear shape of the relationship and the strong decline in risk after about age 65 is striking (Fig. 1). Others have also observed this pattern,¹ but we are not aware of a definitive explanation for the decline. Male gender is also a well-known risk factor for RD. Trauma has been suggested as one explanation for the higher rate in men, but it is also possible that heavy physical work could play a role.

Myopia had the expected strong effect on risk of RD/T, although some previous studies suggest an even stronger risk.³ We think that it is likely that the myopia effect has been somewhat attenuated here because of our proxy measure, which probably contained some nondifferential misclassification. Despite this, we found no evidence that the proxy measure introduced confounding into the associations between RD/T and the other risk factors. The effect of myopia was the one association that appeared to differentiate RD from RT—there was a much stronger association of myopia with RD than RT (Table 2). We are not aware of explanations for this difference.

Previous cataract surgery was a very strong risk factor, as has been previously reported.²¹ In designing the study, we tried to avoid this complicating factor by excluding participants who had cataract surgery within 3 months before their RD/T (or routine eye examination for controls). But the results in Table 2 show clearly that previous cataract surgery (in the same eye) was a strong risk factor even after excluding recent (within 3 months) cases. We did not have complete data on the date of cataract surgery in the records available for abstracting, and so we could not investigate the persistence of this risk over time or evaluate whether or not myopia (or other risk factors) increase this risk. This would be important to investigate further in a future study.

A family history of a close relative with RD also appeared to be a risk factor. It is possible that this might have been affected somewhat by recall bias, with cases more likely to remember or be aware of RD in the family. But a family history risk has also been reported by other investigators.²²

We did not expect to find a negative association of BMI and RD/T, as both previous studies investigating this found positive associations.⁵,⁷ We do not know what might account for this difference, although it may be important to note that the Italian and Swedish populations had much lower distributions of BMI, with few “overweight” or “obese” (BMI >25 and >30, respectively). Among our controls, 69% were overweight or obese versus 7% and 26% in the Swedish and Italian study populations, respectively.

The study by Mattioli et al⁵ mentioned previously was the first epidemiologic study linking heavy lifting to risk of RD. A second epidemiologic study also found evidence that physical exertion, and specifically heavy lifting at work was associated with an increased risk of RD.⁵,²³ The second study was conducted in a cohort of approximately 49,000 Swedish men born in 1949 to 1951, enrolled during their compulsory military service in 1969 to 1970 and followed up to 2009 when they were about 60 years old. Heavy lifting was determined from their occupations in 1990, when they were about 40 years old. Those in occupations requiring lifting of heavy loads (> 20 kg or >44 lbs) at least twice/week had about twice the risk of RD compared to those in the lowest lifting category (relative risk = 2.38, 95% CI 1.15 to 4.93).

In designing the present case–control study to follow up the initial observation of Mattioli, we found it useful to develop explicit hypotheses about the causal mechanisms by which physical exertion might cause RD. In particular, the choice of how to summarize data on exposure—heavy lifting—is dependent on assumptions about the timing of exposure, and whether the effect is cumulative and gradual or acute and rapid.²⁴

It is known that physical exertion causes rapid transient increases in intraocular pressure (IOP), providing a possible link between macro physical activities and intraocular dynamics,^25–27 but how transient increases in IOP might lead to RD is not known. One possible mechanism involves disruption of an evolving posterior vitreous detachment (PVD). RD is most common during the fifth through seventh decades¹ (Fig. 1), when PVD is also occurring in many eyes, and “anomalous” PVD may represent a frequent pathway to RD.¹⁰ During a normal PVD, the vitreous shrinks and its cortex pulls away from the retina without damage. But if the vitreous liquefaction and shrinkage are not synchronized with weakening of vitreoretinal adhesion, a RT may occur.

We hypothesized that short-term peaks in intraocular pressure caused by heavy lifting occurring when an eye is going through PVD can disrupt the normal process, increasing the risk of an RT or RD. A rapid increase in IOP may force apart the vitreous cortex from the retina leading directly to a tear. If this hypothesis is correct, an epidemiologic study should find an association between lifting shortly before the RD and risk. It further follows that risk would be associated with the current job, during or shortly before the event, while exposures in the distant past would not be important, nor would the cumulative exposure over the entire career, except to the extent that a person’s job characteristics may be very similar over her/his career.

The finding that risk of RD is modestly increased among those whose current job involved regular heavy lifting is consistent with this hypothesis, as is the lack of an association between cumulative exposure measures and RD risk. Many (27%) of our participants were retired, and so they had no “current” job. When we restricted the analysis to those under age 65, we found a stronger association between current heavy lifting and RD risk, providing further evidence for the hypothesis. We believe that the stronger association observed in the propensity score matched data was due to this method providing tighter control for the other strong covariates and their interactions than is possible in multiple logistic regression.

There are several potential limitations. First, our study relied on self-reported information for important covariates, including myopia, previous cataract surgery, and history of physical exertion. Regarding the prior hypothesis about the risk from heavy lifting, this study was also limited by the fairly low prevalence of occupational heavy lifting in this population; only 12.2% of the controls and 14.8% of cases reported regularly lifting 30 lbs (13.6 kg) or more in their current job. This is probably due to the types of industries present in the catchment area of central Massachusetts. Future investigations of the role of physical exertion in RD should focus on populations with a higher prevalence of heavy labor. It is possible that cases over-reported their history of physical exertion because of their belief that this may have been the cause of their RD/T. We are doubtful that this might fully explain the findings for two reasons. First, we might expect this recall bias to extend to nonoccupational lifting such as housework, yard work, and hobbies, but we did not find any evidence for an association between these and risk of RD/T. Second, the last item on the survey given to cases was an open-ended question: “Can you think of anything that may have caused your retinal detachment or retinal tear?”. Only 40% of cases provided an opinion, the most common responses were various eye health conditions (11%), and references to heavy exertion or lifting were given by only seven cases (4%).

Caution is warranted in interpreting CIs in Table 3 because there were multiple different exposure variables evaluated. However, as noted above, the exposure metric with the strongest evidence of association corresponded to our primary a priori hypothesis on the timing of exposure and we believe this consistency should be borne in mind when evaluating the findings.

The present study and the two previous studies of lifting and RD risk^5,23 used different measures of occupational heavy lifting exposure. Unfortunately, neither of the two previous studies reported lifting data on the current job. The Italian case–control study by Mattioli et al⁵ found RD risk to be associated with lifetime cumulative lifting exposure,⁵ and the Swedish cohort study by Farioli et al²³ used a measure of occupational lifting for a time point early in participants’ working lives. Our study did not find increased risk of RD/T with measures of lifting designed to be comparable in timing to either of the two previous studies. Thus, further research is needed on the timing of exposure and risk.

This study was not large enough to detect risk from specific nonoccupational activities such as weight lifting, which were reported by only a few subjects. Thus, the lack of evidence for these activities increasing risk may be simply because of insufficient statistical power.

In conclusion, there are potentially important risk factors for RD/T that need further investigation. The possibility that BMI affects risk, whether positively or negatively, has important implications for public health and the contradictory results of this and previous studies should be resolved. The modest elevation in risk we observed among those with current jobs involving heavy lifting requires further study in populations with a higher prevalence of heavy work and more accurate biomechanical exposure data.