*3.2. Study Quality*

The quality of included studies was partly a function of the quality of routine data available, which improved over the years studied. The measures presented also became more meaningful over time; from absolute numbers to rates per 100,000 inhabitants/person-years. Both of these improvements are demonstrated in the chronological presentation of papers in Table 1. For example an early study presented absolute numbers of suicides by age group and gender for 1972 and 1973, but no suicide rates specific to men (or indeed men in specific age groups) [19]. In later studies crude suicide rates were represented as per 100,000 inhabitants [12,22,23] or per 100,000 person-years [18,20,21]. None of the studies used join-point regression models to calculate annual percentage change statistics. Many studies compared suicide rates in different age or gender groups (or between cases and controls) without providing test statistics. Studies investigating risk factors for suicide aggregated genders or age groups, probably for reasons of power, or lacked test statistics.

### *3.3. Risk of Bias across Studies*

The dramatic rise in suicide rates in the early 1970s, increasing from zero in young men aged 15–24 in West Greenland in 1970 to 249 per 100,000 in 1976 [20], may have reflected early under-reporting or misclassification bias. In 1970 all reported suicide cases in West Greenland were for men in the age group 35–59 years of age, but by 1976 suicides were recorded across all age groups above 15 [20]. At the population level, official Greenlandic figures for 1967–1971 indicated an average of 6.4 suicides per year (16.7 per 100,000 Greenlandic-born inhabitants), which had risen to 21.6 per year (53.9 per 100,000) for 1972–1976 [23]. Due to the high relative number of accidents and unidentified causes of death in Greenland, it is possible that a substantial number of suicides are misclassified as accidents [24]. Cultural reasons may underlie the differential recording by age group of suicide *versus* accidental death on the cause of death register, and temporal variations in such practices. The only evidence of selective reporting within studies was for the specific risk factors investigated, which may have reflected inductive bias.

### *3.4. Results of Individual Studies*

### 3.4.1. Aggregated Data for Specific Periods

Over the aggregated period 1970–2011, the highest suicide rates were in men aged 20–24 (at around 410 per 100,000 person-years), with a clear excess of male suicides [18]. Suicide rates for men for the 41 year period 1970–2011 fell with advancing age group, to their lowest level in men aged 65–69 (at around 50 per 100,000 person-years), then increased slightly for men over 70 [18]. These figures are consistent with aggregated suicide data for a shorter 11 year period (1974–1984), showing that suicide rates in young men aged 20–24 were the highest of all age and gender groups in Greenland (at 387 per 100,000 inhabitants per year) compared with 151 per 100,000 for men aged 15–19, 162 per 100,000 for men aged 25–39, and 73 per 100,000 in men over 40 [12]. The gender gap in suicide rates over this period (1974–1984) was greatest in the age group 20–24 (387 for men *versus* 94 for women per 100,000 inhabitants) [12].

Similarly, aggregated suicide data for the shorter 10 year period 1977–1986 show a consistent pattern of highest suicide rates in Greenlandic men aged approximately 20–23, greatly exceeding those for women of the same age [23]. Visual plots presented for average suicide rates per 100,000 population per year showed that suicide risk increased sharply from the ages of approximately 15–17, peaking at almost 600 per 100,000 per year at approximately age 20–23, falling thereafter but with further peaks aged around 35 (approximately 300 per 100,000) and around 55 (approximately 250 per 100,000). A grea<sup>t</sup> disparity between male and female suicide rates was apparent from the age of approximately 17–27 [23].

By the 1990s, aggregated data for the narrower time periods 1990–1999 [21] and 1990–1995 [24] reveal men aged 15–24 to have been the highest risk group. A study presenting visual plots of age-specific suicide rates for the aggregated period 1990–1999 showed suicide rates to be highest in men aged 20–24 in Greenland (approximately 470 per 100,000 person years), closely followed by men aged 15–20 (approximately 430 per 100,000 person years); both considerably higher than those for men or women in other age groups in Greenland, or for men and women of any age in Denmark. This disparity continued up until approximately the age of 45, when rates in men and women in Greenland and Denmark started to converge at around 50 per 100,000 person-years [21].The pattern of suicide risk in Danish men and women over the same period was very different; rising slowly across the age groups but remaining below approximately 70 per 100,000 person-years [21]. In Denmark there was also a much lower ratio of male:female suicides of 1.8 compared to 4.3 for Greenland [21].

Aggregated data for the shortest span of 1990–1995 show that for this period men aged 15–19 had overtaken men aged 20–24 as the group at highest risk, these being the two highest ranking age groups for suicide rates in all age-groups in either gender [24]. Rates for men aged 15–19 during 1990–1995 were approximately 480 per 100,000, declining with age at approximately 440 per 100,999 for men aged 20–24, and 300 per 100,000 for men aged 25–29, and declining thereon. This compared with age-standardised population rates of nearly 110 per 100,000 population over that period. Higher rates for men than women applied in all age groups [24].
