The Effect of Intermittent. Negative Pressure Ventilation and Long-term Oxygen Therapy for Patients With COPD – Methods
The remaining 35 (denoted as group A, date of enrollment from 1984 to 1986) formed the group in which we evaluated the effects of LTOT. In this group, the flow rate of oxygen was individualized in order to maintain the Pa02 value above 60 mm Hg and to avoid harmful changes in PaC02. All patients but one stopped smoking at the beginning of the follow-up. The hemoglobin carbon monoxide values were checked monthly. buy asthma inhalers
Forty-four patients with chronic respiratory insufficiency (group B, date of enrollment from 1976 to 1982) were used as a control group. These patients, who were admitted to our respiratory ICU for ARF, were not prescribed LTOT after discharge. The selection of the patients for the control group comes from the period from 1976 to 1982 where no home treatment with oxygen was prescribed at our institution. Furthermore, only patients in whom the measurements of respiratory function were available were included.
For both groups, the follow-up lasted at least 4 years. All patients were submitted to the same home medical therapy and were seen monthly at our outpatient division. The relapses of ARF during the follow-up of both groups were treated with intermittent negative pressure ventilation (INPV). To obtain an index of the frequency of relapses over time, the number of relapses of each individual patient was divided by the months of survival yielding a ratio denoted as relapses per month (RM). Deaths were directly ascertained from hospital records. Three patients who did not return for 6 consecutive medical visits were considered dead at 1 month after the time of their last control; in these patients, phone calls to family were made to confirm that the patient had died.
To evaluate statistically the trend over time of some variables, the test for trend was used. Accordingly, a regression analysis was ap-plied to the values of the variable measured at different times in each patient thus yielding an individual regression slope for the variable (eg, rate of decline of PaOz over time, rate of increase of PaC02 over time). This statistical analysis has been used by Cooper and Howard.
Survival curves were calculated by actuarial methods. Dicoto-mous variables affecting survival were evaluated by univariate F test statistics (CSS computer program, release 3.0E, StatSoft Inc, USA). Multivariate analysis of survival data was carried out using the proportional hazard model of Cox. The analysis was performed using the step-down method with statistical level set at p < 0.10 (in order to identify variables that influenced survival significantly [p < 0.05] and variables that showed only a statistical trend [0.05 < p < 0.10]). The variables included in the Cox analysis were divided into two groups: (1) variables measured at entry in the follow-up (eg, FEV,, FEVj/VC, Pa02 PaC02, presence of cor pulmonale); and (2) variables measured during the follow-up (eg, number of RM, regression trend of Pa02 and PaCOa, administration of LTOT).
Unless otherwise indicated, all data are presented as mean ± SD. Fisher’s exact test was used to assess statistical significance in 2 X 2 contingency tables. The t test for unpaired samples was used where appropriate.