546 cases. In the process of preparing this tape, it was determined that 21 of these infants were born outside the state of California; these were therefore excluded from the analysis. A total of 525 deaths occurred among the 1968 cohort of live births in California and were included in the final analyses as confirmed sudden unexplained post-neonatal deaths.

The method of analysis included one- and two-way frequency tabulations according to such relevant factors on the death certificate as age, sex, race, and date, place, and cause of death. Birth certificate information, such as date, place, and time of birth, birth order, age of mother and number of previous deliveries to mother, and age, race, and occupation of father was also included.

RESULTS

Confirmation status. Table 2 presents the confirmation status of the 719 deaths reported among the 1968 live birth cohort. For 573 of these cases the response from coroners or medical examiners was ade

quate for confirmation: 45 cases were identified as "sudden death" or "crib death" in the autopsy protocol; 473 cases were confirmed by the coroners or medical examiners to be sudden deaths; 51 cases were identified as not being sudden deaths; and in 4 cases the coroner or medical examiner could not make a decision. A total of 91 cases were determined by the coroner not to be a coroner's case. These were infants who were not included in the definition of sudden unexplained death or who had died in a hospital and/or were under a physician's care and, therefore, did not require a coroner's inquiry. A total of 55 cases required independent review by the panel of UCD pathologists. Of these, 28 cases were confirmed as sudden death, 17 cases were not confirmed, and in 10 cases the UCD pathologists made no decision as to confirmation of sudden death.

Table 3 shows cases confirmed and not confirmed as sudden unexplained deaths, presented according to cause of death on the death certificate. It can be seen that the

173 unconfirmed cases are distributed quite differently according to diagnostic classification than are the confirmed cases. Further, infants with ill-defined or unknown cause of death, interstitial pneumonitis, or thymic hyperplasia recorded on their death certificates were more likely to be classed as sudden unexplained death than were those whose causes of death were recorded as bronchopneumonia, pulmonary congestion, or pulmonary edema. This indicates a degree of variability in the state requirements for identifying a specific cause of death on the certificates for those infants who die suddenly, a finding that requires further investigation or changes in the state requirements for certification of cause of death.

Death rates. As was mentioned earlier, 525 confirmed cases of sudden unexplained death were identified among the 1968 cohort of California live births: a death rate of 1.55 per 1000 live births. A comparison of the California rate with the rates reported by other investigators is shown in table 4. Although there is no uniformity in design and methodology among these studies, the reported death rates, with the possible exception of Seattle (8), Ontario, Canada (9) and Cleveland (10), are somewhat similar. With the exception of the rate reported by Fitzgibbons et al. (3), the death rate in California is the lowest reported in the United States.

Geographic distribution. For purposes of geographic distribution, post-neonatal sudden unexplained death rates by county are grouped into low, medium and high intervals. Counties with fewer than five cases were excluded from analysis. As can be seen in figure 1, there seems to be no pattern in the distribution of high, moderate or low death rates by county in California. For example, four counties (Los Angeles, Orange, San Francisco and Alameda) with very large or dense populations, have low to moderate death rates, while two medium size population counties (San Diego and Sacramento) show high death rates. Further, five counties (Imperial, Merced,

TABLE 3

501

the state, namely, desert, central valley, coastal and mountain. A lack of significant pattern is further demonstrated by a comparison of Orange and San Diego Counties, the two most populous southern coastal counties with temperate climates. These two adjacent counties show respectively a low and high death rate.

Age and season. For several years, it has been the feeling among many investigators that the syndrome of sudden unexplained death is not a distinct epidemiologic entity but rather a reflection of post-neonatal deaths from all causes. This same expression has been reiterated recently by some of the investigators attending a conference on

=

current research findings on infant sudden death (11). We decided, therefore, to compare post-neonatal sudden unexplained deaths with deaths from all other causes according to age at time of death (figure 2). It can be seen that the highest proportion of sudden unexplained deaths occurs before four months of age. During the remaining months of the post-neonatal period the proportion of sudden deaths dramatically decreases to almost 0, whereas the percentage of deaths due to "all other causes" remains at about 5 per cent per month for the remainder of the post-neonatal period. The peak occurrence of post-neonatal sudden unexplained death is found between 28 days to three months of age, with a mean age at death of 2.9 months and a median of 2.4 months. The peak occurrence of deaths due to all other causes is found from the second to the fifth month of age, with a mean of 4.6 months and a median of 3.3 months. Differences in the age distribution of infants who die suddenly and those who die from all other causes appear to be statistically significant (p < .001).

The bimonthly death rates among the cohort of infants born in various two-month periods of 1968 are shown in figure 3. Twomonth intervals were selected to facilitate

RATES/10,000 LIVE BIRTHS

20

FIGURE 3. Post-neonatal death rates by age in months according to cohort month of birth, California, 1968.

comparisons. The peak rate of sudden unexplained death appears to occur between 28 days and three months of age for infants born in September through April. The peak

503

rate for infants born in the months of July and August occurs between three and five months of age, while a peak in death rate for the May and June live birth cohort is not easily detected. The highest bimonthly sudden unexplained death rate is observed among infants born during September and October. Infants born in November and December, or May and June, appear to be at a generally lower risk of sudden unexplained death than those born in January and February or March and April.

The coincidence of critical age (1-3 months) and season (winter) appears to heighten the risk of sudden unexplained death. Infants born in September or October reach the winter season (NovemberJanuary) at the critical age of one to three months placing them at a much higher risk than, for instance, that risk reflected by the rates of death for infants born in November or December. For all month-of-birth cohorts, except July and August, peak age of death is one to three months. However, a shift in peak age of death for the July August live birth cohort to three to five months is observed. While these infants born in July and August are two to three months older than the September-October live birth cohort when they reach the colder months, it appears that they remain at high risk of death due to the impact of season. The colder, early winter months apparently exaggerate the already high risk of death for those infants one to five months of age, that is, infants born in September-October, or July-August.

Rates of post-neonatal death due to all other causes by age in months according to bimonthly live birth cohort are shown also in figure 3. The peak age of occurrence of sudden unexplained death appears to be different from the peak age of death due to all other causes. That is, except for the March and April live birth cohort, there appears to be an extended period of peak occurrence of deaths due to all other causes. Also, while death rates due to sudden unexplained causes appear generally to decline

dramatically after four months of age, the decline in death rate due to all other causes appears noticeably protracted during the latter months of the post-neonatal period. Finally, whereas the highest rates of postneonatal sudden unexplained death occur among the September and October live birth cohort, the highest rate of death due to all other causes is found among the November and December live birth cohort.

The relative distribution of post-neonatal deaths (sudden and all other causes) by month of occurrence is shown in figure 4. The proportionate distribution of sudden unexplained deaths by month was found to be significantly different (p < .001) from that expected on the basis of the percentage distribution of deaths due to all other causes. Many other investigators (1, 12) report a high proportion of deaths in winter months even after adjustment for variation of number of births by month. While findings in the present study are in general agreement with these reports there appears to be an aggregation of deaths early in the winter period (November and December) and fewer deaths are seen in the months of January, February and March as observed in one report (12).

It appears that the overall pattern of distribution of rates between the two groups of infants is not very similar. This suggests that perhaps those infants who die suddently may not be part of the cohort of infants who die from all other causes.

Sex. It is a common belief that sudden and unexplained death is more frequent in male than female infants (2). Many investigators have reported a marked sex differential in the frequency of this syndromean average of about 70 per cent of all reported cases are males. The rate of post-neonatal sudden unexplained death in California is 1.82 per 1000 live births among male infants as compared with 1.26 per 1000 live births in female infants. The observed excess in deaths for males was statistically significant (p < .001) from that expected on the basis of the distribution of live