Re-evaluation of acid-base prediction rules in patients with chronic respiratory acidosis

1Respiratory Epidemiology Unit, Montreal Chest Institute, McGill University, Montreal, Quebec; 2Department of Medicine, Duke University Medical Center, Durham, North Carolina, USA Correspondence and reprints: Dr Sandra Dial, Montreal Chest Institute, 3650 St Urbain, Room K1.14, Montreal, Quebec H2X 2P4. Telephone 514-842-1231 ext 32336, fax 514-843-2083, e-mail sandra.dial@mcgill.ca T Martinu, D Menzies, S Dial. Re-evaluation of acid-base prediction rules in patients with chronic respiratory acidosis. Can Respir J 2003;10(6):311-315.

H ydrogen ion concentration is a critical determinant of many physiological functions (1).For this reason, the blood pH is very tightly regulated and normally ranges between 7.38 and 7.42 in humans and many other species (2).Many physiological systems are involved in this regulation, including the respiratory system and kidneys, as well as red blood cells, proteins and the bicarbonate buffering system within the blood.
The current prediction equations used to assess the acidbase status in patients with chronic respiratory acidosis were derived from experiments in dogs (3)(4)(5).In these experiments, large changes (greater than 20 mmHg) in the partial pressure of carbon dioxide (PCO 2 ) were induced and sustained for a maximum of seven days.However, this is not the typical rate of increase, nor duration, of hypercapnia in patients with slowly progressive diseases such as chronic obstructive pulmonary disease (COPD).In these patients, normal or near-normal pH values have been observed (6).These are often attributed to the presence of a metabolic alkalosis due to diuretics and/or corticosteroids, superimposed on the primary respiratory acidosis.In four earlier studies (7)(8)(9)(10), the degree of acid-base compensation was greater in patients with chronic respiratory acidosis than that described in the canine experiments.However, the patients studied were hospitalized and clinical stability was not defined (10), or was defined as elevated PCO 2 that was stable for only three days (7)(8)(9).It is possible that, as in the dog experiments, the measurements were made before there was complete acidbase compensation.
The present study was performed to re-evaluate the relationships between the arterial pH, PCO 2 and bicarbonate levels in patients with chronic respiratory acidosis who were clinically stable for at least one month.

METHODS
Patients were considered eligible if they had arterial blood gas analyses performed as outpatients at the Montreal Chest Institute, Montreal, Quebec, had COPD or cystic fibrosis (CF) and were hypercapnic (arterial PCO 2 of 45 mmHg or higher).Patients were excluded if they had taken diuretics, oral steroids or angiotensin-converting enzyme inhibitors, or had renal failure in the month before the arterial blood gas sampling.This study was approved by the Ethics Committee of the Montreal Chest Institute.
Eligible patients were defined as clinically stable at the time of the arterial blood gas sampling if they met the following criteria: the arterial blood gas analysis was performed during a routinely scheduled outpatient visit; the blood gas analysis was done to assess need for home oxygen or rehabilitation; there were no new respiratory symptoms or change in medications for at least one month before and after the blood gas analysis; and they were considered stable by their primary physician.
Arterial blood gas results were also extracted from patients who were eligible but not clinically stable at the time, as defined above.These patients were defined as clinically unstable and their blood gas analyses were used as observations for comparison with the observations from clinically stable patients.

Data collection
Data abstracted retrospectively from the patients' charts included age, sex, comorbid diseases, medications, diagnosis, use of home oxygen, baseline pulmonary function tests and all results of outpatient arterial blood gas measurements.Clinical stability was assessed from the data collected by one of the investigators, without knowledge of the arterial blood gas results.
The arterial blood gas measurements were analyzed using a Bayer 800 series analyzer (Bayer, Canada).This instrument provides linear measurements for the PCO 2 in the range of 5 mmHg to 250 mmHg (machine specification).To ensure accuracy of the measurements, the machines are calibrated using three controls on a daily basis; the results of quality assessments over the study period are listed below.The within-run precision was within 2% at all levels.The bicarbonate levels were calculated from the arterial PCO 2 and pH using the Henderson-Hasselbalch equation.

Data analysis
To estimate the relationship between PCO 2 and arterial pH, bicarbonate and hydrogen ion concentrations by comparison with other studies, linear regression was performed using SAS (SAS Institute, USA).

RESULTS
Arterial blood gas analyses were performed on 18 eligible patients who were deemed to be clinically stable at the time of sampling.As shown in Table 1, four patients had CF and 14 had COPD.Although these two groups were very different in terms of age, their arterial blood gas results and acid-base compensations were very similar.Therefore, results from patients with these two diagnoses were combined in all tables and figures.The overall average pH was 7.40 and PCO 2 was 58 mmHg.The lowest observed pH was 7.37; a pH lower than 7.38 was measured in only three instances (17%) (Table 2).
The slope of the relationship between arterial PCO 2 and pH estimated from regression is shown in Figure 1.Similarly, the estimated relationship between arterial PCO 2 and bicarbonate is shown in Figure 2. As shown in Table 3, an increase of 10 mmHg in the PCO 2 was associated with a decrease of only 0.014 in the pH and an increase of 5.1 mmol/L in the bicarbonate level.Using the regression coefficients and intercept, in a stable patient with a PCO 2 of 55 mmHg, one could predict that the pH would be 7.40 with a bicarbonate level of 33 mmol/L.There were 17 blood gas samples taken from 17 eligible patients who did not meet the criteria of clinical stability at the time of arterial blood gas sampling.A test performed to evaluate whether the two regression curves were statistically significantly different almost reached statistical significance (P=0.06).The estimated relationships between arterial pH, bicarbonate and PCO 2 in the unstable group were very similar to those obtained  in the canine experiments and in studies of patients whose clinical stability was defined in an inpatient setting (Table 4).

DISCUSSION
In our study, 18 patients with chronic hypercapnic respiratory disease and without any possible metabolic alkalosis underwent arterial blood gases analyses when they had been without any change in respiratory symptoms or medications for at least one month.Despite having PCO 2 levels as high as 77 mmHg, all of the patients maintained their pH at 7.37 or greater, and more than 80% had a pH of 7.38 or greater.In patients with chronic hypercapnia, acid-base compensation appears more complete than previously believed.
The limitations of our study include the retrospective design, which can result in incomplete data or confounding, because blood gases are often taken to evaluate symptomatic illnesses (ie, when patients are not stable).Incomplete data was not a problem in the patients reviewed, who form part of a large, prospectively gathered patient database at the Montreal Chest Institute.As well, the indications for blood gas analysis were assessed for home oxygen or rehabilitation programs, both of which emphasize that patients should be stable at the time of the blood gas analysis.The major limitation was the small number of observations.However, statistically significant associations were nevertheless detected, which suggests that the relationships were strongly correlated, because they were detected even with limited power.The small numbers in the study resulted in wide CIs obtained around the estimates of the slope and intercept.
The major strength of our study is that patients were carefully defined as clinically stable for at least one month.As well, patients with disorders likely to cause a superimposed metabolic alkalosis were excluded.Therefore, the findings are of much greater relevance to clinical practice than those from an experimental dog model (3,4) or short term observations in hospitalized patients (7,10).These studies are likely to have underestimated the compensation that patients with chronic respiratory failure may achieve.
The rules for the expected compensation in the setting of acute and chronic respiratory acidosis, which are still applied   today (11,12), were developed in 1965 from experiments in dogs (3,4).In one of these canine studies (3), 10 dogs were exposed to normal atmosphere and then to successive concentrations of 7%, 11% and 17% carbon dioxide, each for five to seven days.A chronic state was defined as a sustained elevation in the arterial PCO 2 for four to five days.This definition was based on previous observations that after four to five days of exposure to a carbon dioxide concentration of 12%, the correction of the pH appeared to reach a plateau (5).How applicable are the results of this experiment to patients with slowly progressive lung diseases?First, the experimental procedure of three episodes of acute hypercapnia, each maintained for less than one week, is very different from patients with chronic lung diseases, who develop hypercapnia over a period of years.Also, in the canine study, hypernatremia and significant bicarbonate losses were described in all of the dogs who survived the experiment; however, hypernatremia is not a usual finding in patients with hypercapnia from chronic respiratory failure.
As summarized in Table 3, the relationships between PCO 2 , bicarbonate and pH found in the unstable patients in this study were more comparable to the classic dog experiments (3)(4)(5) and two studies in humans (10,13).In one of these human studies, 20 patients were evaluated after three days of stability while hospitalized for a respiratory decompensation (10).In the other study, 420 patients with chronic lung disease were evaluated, but chronicity of the respiratory acidosis, patient medications and other confounding factors were not described (13).In a third study among hospitalized hypercapnic patients whose PCO 2 levels varied less than 10% over three days (9), a normal arterial pH was seen in 87% of the patients with PCO 2 between 46 mmHg and 55 mmHg, compared with only 50% of patients with PCO 2 between 56 mmHg and 65 mmHg and 32% of patients with PCO 2 between 66 mmHg and 75 mmHg.The results among the stable patients in this study are most similar to the results of a study examining 247 arterial blood gas samples in 106 outpatients (8).In that study, although pH was not reported, for each increase of 10 mmHg in the PCO 2 , the bicarbonate level increased by 5.3 mmol/L -very similar to the findings among stable patients in the present study.Ingram et al ( 14) studied patients with chronic hypercapnia who had acute changes in their PCO 2 , with the aim of trying to define the relationship between the change in pH and increases in the PCO2 at different baseline PCO2 levels.The demographics of the patients in the steady state were not provided, but they reported the regression equation of the correlation between the steady state hydrogen ion concentration and PCO 2 in the patients studied.The correlation was very similar to the one found in the present study.In a recent report (15) examining the strong ion difference in patients with chronic hypercapnia, while not specifically examining the correlation of pH and PCO 2 , the authors found that instead of a predicted pH of 7.355, they found a measured pH of 7.372 in a group of 12 patients with a mean PCO 2 of 54 mmHg.
There is a physiological concept that compensation should never be complete, because then the stimulus to compensate would be lost.This would mean that as the pH approaches normality, the kidney would no longer be stimulated to retain   bicarbonate and the pH would reach equilibrium below the normal range.However, in animal studies, the signal to compensate was not the low pH.Instead, the elevated PCO 2 directly stimulated the proximal tubule of the kidney to increase bicarbonate reabsorption (16)(17)(18).This implies that increased reabsorption of bicarbonate should continue as long as the PCO 2 remains elevated, raising interesting questions regarding feedback control of this compensatory response.Chronic adaptation is believed to be entirely due to renal adaptive processes, with increased net acid secretion, primarily in the form of ammonium, reduced reabsorption of chloride and increased reabsorption of bicarbonate ion (16,18).One study found changes in the gastrointestinal handling of chloride ion but little change in renal chloride excretion (19).

CONCLUSIONS
Although the mechanisms are not understood, renal compensatory mechanisms in chronic respiratory failure appear to be more complete than previously recognized.The clinical implication is important.In a patient with chronic hypercapnia, an arterial pH below the normal range should suggest the occurrence of worsening respiratory acidosis or new metabolic acidosis.

Figure 1 )
Figure1) Relationship between pH and partial pressure of carbon dioxide (PCO 2 ) in stable hypercapnic outpatients (estimates for slope regression in 18 patients, r 2 = 0.39)

TABLE 4
Comparison from the present and earlier studies of regression equations of the relationship of chronic steady state partial pressure of carbon dioxide (PCO 2 ) to hydrogen ion concentration [H+], and predicted bicarbonate concentrations and pH at two levels of PCO 2