Per-capita claims rates for decompression sickness among insured Divers Alert Network members

UHM 2012, Vol. 39, No. 3 DCS AMONG INSURED DAN MEMBERS Per-capita claims rates for decompression sickness among insured Divers Alert Network members Petar J. Denoble M.D. 1, Shabbar I. Ranapurwala M.P.H. 1,2, Panchabi Vaithiyanathan Ph.D. 2, Richard E. Clarke 3, Richard D. Vann Ph.D. 2 1 Medical Research, Divers Alert Network, Durham, N.C. USA 2 Department of Epidemiology, Gillings School of Global Public Health, UNC, Chapel Hill, N.C. USA 3 National Baromedical Services, Columbia, S.C. USA CORRESPONDING AUTHOR: Dr. Petar Denoble pdenoble@dan.org ABSTRACT Decompression sickness (DCS) in recreational diving 20.5 per 10,000 member-years. Based on the ageadjusted DCR, males submitted 28% more claims than is a rare and usually self-limiting injury, but permanent disability can occur. Incidence rate estimates are difficult to establish because the number of divers at risk is tween 35 and 40 years of age and disappeared by the females. Male-to-female difference was greatest be- usually unknown in population samples with well-documented DCS. We estimated the annual per-capita DCS year age category, after which DCR declined with in- mid-50s. Highest rates were observed in the 30- to 39- incidence rates for 2000-2007 based on insurance claims creasing age. Highest yearly DCR was estimated in submitted by members of the Divers Alert Network 2002. Insurance dropout rate was greater among those (DAN), Durham, N.C., with dive accident insurance. who had DCS in the first year of their insurance compared to those who did not have DCS in their first year. The overall per-capita DCS claims rate (DCR) was INTRODUCTION Decompression sickness (DCS) can occur after a reduction in barometric pressure following compressed-gas diving and during air travel or extravehicular activity [1]. Its effects are often acute and self-limiting, but early recompression while breathing 100% oxygen is recommended to reduce the chances of residual injury. Populations at risk include astronauts, aviation air crew and military, scientific, commercial and recreational divers. Recreational divers are the largest group, numbering in the millions [2], and about 1,000 are treated for DCS annually [3, 4]. Reports of DCS in well-defined diving populations are rare, so credible estimates of DCS incidence rates are difficult to find. We estimated the annual percapita incidence rates among members of the Divers Alert Network (DAN) who purchased dive accident insurance over an eight-year period from 2000-2007. Insured DAN members are a distinct subpopulation in which sex, age and injury claims history are known. Our measure of DCS cases was based on submitted injury claims. DCS incidence rates are commonly calculated as the number of cases divided by the number of dives, where dives are the measure of exposure [1]. We did not know how many dives each insured DAN member had made, so we reported the number of claims per diver per year. A per-dive metric is a more useful measure to quantify exposure for event-based incidence rates, but in its absence, our per-diver metric nonetheless allowed us to investigate the distribution of rates within our population. We calculated the annual per-capita DCS claims rates (DCR) and assessed the influences of sex, age and multiple DCS claims. METHODS De-identified injury claims data for the period 2000-2007 were provided by the claims adjudicator, Palmetto Health of Columbia, S.C., who verified the diagnosis. The selection criterion for a claim was the ICD-9 code 993.3, assigned and verified by adjudicator based on diagnosis provided by the treating physician. The ICD-9 codes define 993 as effects of air pressure, while 993.3 is mentioned as caisson disease and Copyright 2012 Undersea & Hyperbaric Medical Society, Inc. 709

UHM 2012, Vol. 39, No. 3 DCS AMONG INSURED DAN MEMBERS further elaborated as bends, decompression sickness, compressed-air disease and divers palsy or paralysis. Besides the ICD code, claims data included a randomly generated identifier for diver, date of injury, sex, age of claimant and first year of purchasing DAN insurance. The number of divers at risk for 2000-2007 was drawn from divers who had purchased DAN dive accident insurance. De-identified membership data included age, sex and the year when the diver purchased DAN insurance. The study was approved by the Duke University Medical Center Institutional Review Board. The resultant data set was such that each year of participation for any member contributed to one observation in the data set. This means that if one individual was in the dataset every year from 2000 to 2007, this member would contribute to eight observations. Hence, the total observations in the dataset were equal to the total exposed member-years. The DCR was calculated as the number of claims divided by the number of insured member-years and was reported per 10,000 insured member-years. Insured members were grouped into seven age categories: 15-19, 20-29, 30-39, 40-49, 50-59, 60-69 and >70 years. DCR was calculated for each category and compared with the referent category, 40-49 years, which was the largest group; we calculated rate ratios (RR) and rate differences (RD). (RR is the ratio of DCR of each age group and the referent, while RD is the difference between the DCR for each age group and the referent). The age-adjusted Mantel-Haenzel s rate ratios (RR mh ) and rate differences (RD mh ) among males and females were calculated by Mantel-Haenszel s (MH) pooling using epi-sheet [5]. The influence of age and sex on DCR was assessed by logistic regression with a dichotomous outcome claims variable (DCS claim = Yes or No). Effect measures were reported with 95% confidence limits where applicable. Logistic regression also assessed the influence of calendar year on DCR, with 2000 as the referent year. To investigate the influence of submitting a DCS claim on a diver s inclination to drop out of dive insurance, we compared divers who submitted a claim in their first insured year with those who submitted in subsequent insured years. Divers who bought insurance for the first time during 2000-2003 and submitted a claim in the first year were assigned to a group called 1st Year Claims. During the same period, divers who did not submit a claim in their first year were assigned to a group called Non-1st Year Claims. We followed both groups for the next four years and calculated their respective survival rates. Survival rate was defined as the percentage of individuals who renewed their insurance in the followup period with respect to the previous year of insurance. We defined dropout rate as the percentage of individuals who did not renew membership from the previous year, mathematically represented as 1 Survival Rate. The trends of survival rates from both the groups were compared using a Mantel-Cox chi-square statistic. We also estimated the crude relative risk of future claims between these groups to assess whether the surviving members of the groups differed in claims submission. For dropout analysis, we used DAN insured members who enrolled for insurance for the first time between years 2000 to 2004. We did not use members who enrolled for the first time prior to or after this period. RESULTS There were 1,304,358 insured member-years and 2,672 claims over the eight-year period, with a crude DCR of 20.5 per 10,000 insured member-years [95% confidence interval: 19.7, 21.3]. Males contributed 64% of member-years to our data set, while females contributed 36%. DCR for males was 22.1 [95% confidence interval: 21.1, 23.1] and was 17.6 [95% confidence interval: 16.4, 18.8] for females. Known ages of claimants ranged from 15-80 years, but age was missing for 26,182 insured members (2.01%) and nine claimants (0.3%). The age-adjusted male-to-female MH rate ratio (RR mh ) was 1.28 [95% confidence interval: 1.19, 1.39], indicating that, overall, males submitted 28% more claims than females. The age-adjusted rate difference (RD mh ) between males and females was 5.79 per 10,000 members [95% confidence interval: 4.34, 7.24], indicating that males submitted nearly six more claims than females per 10,000 member-years. The P-homogeneity was 0.042 for the age-adjusted RR mh and 0.0008 for the ageadjusted RD mh, indicating that all the seven age categories were heterogeneous. Figure 1 (facing page) shows the relationship of DCR to age for males and females. The best logistic regression fit was achieved with a quadratic transformation of age. For both sexes, DCR increased until age 40, then decreased at greater ages. Rates for males were generally greater than for females with the largest difference between ages 35 and 40 and no difference after the mid-50s. Table 1 (facing page) shows the age-specific DCR estimates for each age group and the corresponding RRs and RDs in comparison to the 40-49 years age reference group which comprised 29.5% of the membership. The highest DCR of 26.6 DCS claims per 10,000 710 P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann

FIGURE 1 Male and female DCS claims vs. age DCS claims rate / 10,000 member-years Age of DAN member Figure 1 Male and female decompression sickness claim rates vs. age in insured Divers Alert Network members from 2000-2007 males;... females; 95% confi dence limits TABLE 1 Age-specifi c estimates and confi dence intervals for the total DAN member population from 2000-2007 Age Member Cases DCR 95% Rate 95% Rate 95% group# years confi dence ratio confi dence difference confi dence intervals intervals intervals <20 60,984 33 5.41 3.56, 7.26 0.27 0.19, 0.38-14.69-17.01, -12.36 20-29 136,970 346 25.26 22.60, 27.91 1.26 1.11, 1.43 5.16 2.15, 8.17 30-39 281,994 750 26.60 24.69, 28.49 1.32 1.20, 1.46 6.49 4.12, 8.86 40-49 385,017 774 20.10 18.68, 21.51 1.00 0.00 50-59 314,425 614 19.53 18.00, 21.07 0.97 0.87, 1.08-0.58-2.68, 1.52 60-69 85,674 135 15.76 13.10, 18.41 0.78 0.65, 0.94-4.35-7.35, -1.33 >70 13,112 11 8.39 3.43, 13.34 0.42 0.23, 0.76-11.71-16.86, -6.56 Missing 26,182 9 Total 1,304,358 2,672 20.49 19.49, 21.50 DCR = decompression sickness claims rate; Cases = number of decompression sickness claims = number / 10,000 insured DAN member-years over an eight-year period; # = in years member-years [95% confidence interval: 24.7, 28.7] was in the 30-39 age group with a relative risk of 1.32 [95% confidence interval: 1.20, 1.46] and a rate difference of 6.49 DCS claims per 10,000 member-years [95% confidence interval: 4.12, 8.86]. Figure 1 and Table 1 indicate that the DCR increased in younger divers until reaching a maximum at ages 35-45 and declined progressively thereafter for both males and females. The mean age of DAN members increased every year from 2000 to 20007, as represented in Figure 2 (Page 712) suggesting that the DAN scuba diving population aged over the eight-year period. P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann 711

FIGURE 2 Mean age Calendar year Figure 2 Mean age of insured Divers Alert Network members from 2000-2007 FIGURE 3 DCS claims DCS claims rate / 10,000 member-years Mean age Calendar year Figure 3 Decompression sickness claims rates of insured Divers Alert Network Members versus calendar years (2000-2007) and 95% confi dence intervals The yearly DCRs among insured DAN members are shown in Figure 3 (above). The highest DCR was estimated in 2002 at 25.7 [95% confidence interval: 23.3, 28.2] and the lowest in 2006 at 14.9 [95% confidence interval: 13.2, 16.9]. Regression analysis showed that the aging diving population (Figure 2) did not change the yearly DCR estimates. Figure 4 (facing page) shows the trends of survival rates of the 1st Year claims and Non-1st Year claims groups. over the four-year period, the survival rate in the Non-1st Year group was higher than in the 1st Year group (Mantel-Cox Chi-square = 57.59, DF=1, p < 0.0001 ) indicating that divers who submitted claims in their first insured year had a higher dropout rate than 712 P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann

FIGURE 4 Survival rates in renewing members Survival rates Year of insurance Figure 4 Survival rates of renewing insurance plan in the 1st Year and Non-1st Year claims groups of insured Divers Alert Network nembers for a four year period between the years 2000-2007 divers who did not submit claims in their first insured year. The rate of claiming DCS for a four-year followup period was the same in 1st Year Group (1.36%) as in the Non-1st Year Group (1.60%), with a RR of 0.84 [95% confidence interval: 0.49, 1.48]. Thus, divers were less likely to renew their insurance if they submitted a DCS claim in the first year even though the two groups were equally likely to submit claims in subsequent years. DISCUSSION We calculated per-capita DCS claims rate (DCR) as a surrogate measure of DCS incidence rate for insured DAN members by including all insured members and all claims that they submitted in the eight-year period. We found a crude DCR of 20.5 claims per 10,000 insured members, with a peak DCR in the 30-39 age range and a lower DCR for women (Figure 1). The mean age of insured DAN members increased by about 0.5 years per calendar year, from 41 to 44.3 years (Figure 2), but this increase in age was not associated with the DCR for each calendar year (Figure 3). The highest DCR occurred in 2001. Divers who submitted claims in the first insured year were more likely to drop out than divers who submitted claims in later years. This suggests that treatment for DCI early in one s diving career may be a disincentive to continue diving despite similar DCS risks in the subsequent four years (Figure 4). By using measured rather than estimated denominator and numerator, we avoided possible sampling errors. The only potential comparison we could find was from the British Sub Aqua Club (BSAC), who reported membership data and DCS incidents for 2000-2007 from which we estimated a higher per-capita DCS rate of 32.59 incidents per 10,000 member years [6]. However, we do not know if the reported incidents were serious enough to require therapy or whether insurance claims were filed. On the other hand, BSAC divers were reported to make an average of 37 dives per year [6] while a 1990 survey of DAN divers reported only 25 dives per year [7]. If these numbers are applicable to the per-capita DCS rates, the per-dive rates are closer at 0.88 cases per 10,000 dives for BSAC and 0.82 cases per 10,000 dives for insured DAN members. older divers have been suspected to be at greater DCS risk [8-10], and several studies reported increased post-dive venous gas emboli (VGE) grade with increasing age [12, 13], but we found the per-capita claims rates to peak in age group 30-39 and to decrease for older divers. This does not necessarily mean decreased DCS susceptibility with age, as older people may dive more conservatively. The higher DCR for males over females between ages 20-50 years may also have been behavioral rather than indicating greater intrinsic risk, as some reports found no differences between the genders (10, 14). P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann 713

FIGURE 5 DCS claims vs. death rates Per 10,000 member-years Age (years) Figure 5 Comparison of DCS claims rates and death rates [15] from the same insured DAN membership population Figure 5 (above) shows a comparison between percapita DCS claims rates and diving fatalities [15] taken from the same insured DAN membership population. The mean claim rate was about 10 times greater than the death rate. Males had significantly higher claims and death rates than females, but the death rate rose progressively with age rather than decreasing after a middle-age peak. The increasing death rate with age was associated with a greater occurrence of cardiac events [15], while we surmise that the decrease with age for DCS claims was a result of more conservative diving by both males and females. LIMITATIONS Using insurance claims as a surrogate for DCS is subject to error. Divers with symptoms consistent with DCS are encouraged seek evaluation at a recompression facility, and if DCS is suspected, a test of pressure is common. Should the test lead to a diagnosis of DCS being rejected, however, the claim might still be assigned an ICD-9 Code representing DCS in order to cover the costs. on the other hand, untreated ambiguous DCS cases would not be included in these data. DAN Members who purchased dive accident insurance may not be representative of all recreational divers, and generalizations based on the findings of this paper may be inappropriate for other populations. Per-dive incidence rates are a better measure of exposure than per-diver rates, but we did not know how many dives each claimant and insurance holder had made, and therefore, could not compute the per-dive rates. 714 P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann

UHM 2012, Vol. 39, No. 3 DCS AMONG INSURED DAN MEMBERS REFERENCES 1. Vann RD, Butler F., Mitchell SJ, Moon RE. Decompression illness. Lancet. 2011; 377, 153-64. 2. U.S. Census Bureau. Parks, Recreation, and Travel: Participation in selected sports activities: 1998, Table No. 435; Statistical Abstract of the United States: 2000. 3. Pollock N, Dunford R, Denoble P et al. Annual diving report. Durham, NC, United States : Divers Alert Network, 2008. 4. Vann R, Uguccioni D, Freiberger, J et al. Annual diving report. Durham, NC, United States: Divers Alert Network. 2006. 5. Rothman K. Episheet: Spreadsheets for the analysis of epidemiologic data. May 21, 2007. 6. Cumming B. National Diving Committee Diving Incident Report. Cheshire, United Kingdom : The British Sub-Aquatic Club, 2000-2007. 7. Dear GdeL, Uguccioni DM, Dovenbarger JA, Thalmann ED, Cudahy E, Hanson E. Estimated DCI incidence in a select group of recreational divers. Undersea Hyperb Med. 1999; 26(suppl): 19. 8. Gray JS. 1951. Constitutional factors affecting susceptibility to decompression sickness. In Decompression sickness, ed. J. F. Fulton, 182-191. Philadelphia: Saunders. 9. Bradley ME. Metabolic considerations in fitness to dive. Undersea and Hyperbaric Medical Society. 1987; 98-103. 10. Hagberg M, Ornhagen H. Incidence and risk factors for symptoms of decompression sickness among male and female dive masters and instructors--a retrospective cohort study. Undersea Hyperb Med. 2003; 30, 93-102. 11. Webb JT, Kannan N, Pilmanis AA. Gender not a factor for altitude decompression sickness risk. Aviat Space Environ Med. 2003; 74, 2-10. 12. Dunford RG, Vann RD, Gerth WA, Pieper CF, Huggins K, Wachholz C, Bennett PB. The incidence of venous gas emboli in recreational divers. Undersea Hyperb Med. 2002; 29(4): 247-259. 13. Carturan D, Boussuges A, Vanuxem P, Bar-Hen A, Burnet H, Gardette B. Ascent rate, age, maximal oxygen uptake, adiposity, and circulating venous bubbles after diving. Aviat Space Environ Med. 2009 Jan ;80 (1):15-9. 14. Zwingelberg KM, Knight MA, Biles JB. Decompression sickness in women divers. Undersea Biomed Res. 1987; 14, 311-7. 15. Denoble, P, Pollock N, Vaithiyanathan P. et al. Scuba injury death rate among insured DAN members. Diving and Hyperbaric Medicine. 2008; 38, 182-188. u P.J. Denoble, S.I. Ranapurwala, P. Vaithiyanathan, R.E. Clarke, R.D. Vann 715