Nine Washington area hospitals had more deaths than expected and three had markedly fewer in the government's unprecedented look at 1984 Medicare overall death rates, as well as death rates in nine illnesses or operations.

The government and these hospitals' administrators and doctors pointed out that any single hospital's results don't necessarily prove either competence or incompetence. Unfavorable numbers could merely mean a hospital had sicker patients, more emergency patients or even a run of statistical tough luck.

The good numbers, conversely, may just mean getting more affluent, healthier patients who are better prospects for survival -- or, again, having a lucky run.

The government compiled the data from hospital billing figures, not always the most accurate source. It derived each hospital's "expected" death rate by adjusting the figures for patients' age, sex and race (race because a high proportion of black or Hispanic patients would generally mean poorer, sicker ones), and for the area's average. But it did not consider severity of illness or several other possible factors that can affect death rates. For example, a hospital like D.C. General or a rural hospital that has trouble finding nursing home beds for the chronically ill -- and must hang onto them longer -- might for that reason alone have some high death rates.

A hospital like Sibley in Northwest Washington, as a Sibley doctor said, might look good because it transfers some of its sickest patients to a larger medical center.

Given such caveats in assessing these numbers -- but remembering too that they will at least make many hospitals look at themselves very hard -- the government reported that on the plus side:

*Georgetown University Hospital might have expected an overall 5.3 percent death rate; its actual rate was 1.9 percent.

*Sibley Hospital might have expected an overall 4.8 percent death rate; its rate was 2.9. Sibley also might have expected 23.4 percent of its acute myocardial infarction (heart attack) patients to die; only 8.9 percent died.

*Washington Adventist Hospital might have expected 22 percent of its myocardial infarction patients to die; 7.1 percent died.

On the less favorable side, the expected and actual death rates in specific categories for nine hospitals are shown in the table at right.

Representatives of several hospitals said the adverse conclusions fail to reflect how ill their patients were. For example, from Southern Maryland: "We get very sick senior citizens from all over southern Maryland. We take all emergencies."

From Greater Southeast: "Many of our patients have other complications and preexisting problems. The socioeconomics of our area means we have a much sicker population."

Some hospitals said their patient totals were too small for any conclusions. But the government said it included only statistically significant numbers, meaning that in no more than five cases in 100 should the outcome be a result of mere chance.

The hospitals generally said they regularly examine all deaths and any apparent excess death rates. Still, several said they will now investigate further.