Clin Res Cardiol DOI 10.1007/s00392-015-0850-3
ORIGINAL PAPER
Biological variation, reference change value (RCV) and minimal important difference (MID) of inspiratory muscle strength (PImax) in patients with stable chronic heart failure Tobias Ta¨ger1 • Miriam Schell1 • Rita Cebola1 • Hanna Fro¨hlich1 • Andreas Do¨sch1 Jennifer Franke1 • Hugo A. Katus1 • Frank H. Wians Jr.2 • Lutz Frankenstein1
•
Received: 6 August 2014 / Accepted: 27 March 2015 Ó Springer-Verlag Berlin Heidelberg 2015
Abstract Background Despite the widespread application of measurements of respiratory muscle force (PImax) in clinical trials there is no data on biological variation, reference change value (RCV), or the minimal important difference (MID) for PImax irrespective of the target cohort. We addressed this issue for patients with chronic stable heart failure. Methods and results From the outpatients’ clinic of the University of Heidelberg we retrospectively selected three groups of patients with stable systolic chronic heart failure (CHF). Each group had two measurements of PImax: 90 days apart in Group A (n = 25), 180 days apart in Group B (n = 93), and 365 days apart in Group C (n = 184). Stability was defined as (a) no change in NYHA class between visits and (b) absence of cardiac decompensation 3 months prior, during, and 3 months after measurements. For each group, we determined within-subject (CVI), between-subject (CVG), and total (CVT) coefficient of variation (CV), the index of individuality (II), RCV, reliability coefficient, and MID of PImax. CVT was 8.7, 7.5, and 6.9 % for groups A, B, and C, respectively. The II and RCV were 0.21, 0.20, 0.16 and 13.6, 11.6, 10.8 %, respectively. The reliability coefficient and MID were 0.83, 0.87, 0.88 and 1.44, 1.06, 1.12 kPa, respectively. Results were similar between age, gender, and aetiology subgroups.
& Lutz Frankenstein [email protected] 1
Department of Cardiology, Angiology, and Pulmonology, University of Heidelberg, Im Neuenheimer Feld 410, 69120 Heidelberg, Germany
2
Department of Pathology, Baylor University Medical Center, Dallas, TX, USA
Conclusion In patients with stable CHF, measurements of PImax are highly stable for intervals up to 1 year. The low values for II suggest that evaluation of change in PImax should be performed on an individual (per patient) basis. Individually significant change can be assumed beyond 14 % (RCV) or 1.12 kPa (MID). Keywords Chronic heart failure Inspiratory muscle strength PImax Reference change value Minimal important difference Biological variation
Introduction Respiratory muscle dysfunction is an important co-morbid condition in patients with chronic heart failure (CHF). It presents both in the form of reduced respiratory muscle force [1–4] and reduced respiratory muscle endurance [5, 6]. Respiratory muscle force is estimated by measuring maximum inspiratory mouth occlusion pressure (PImax) [7]. In CHF patients, respiratory muscle dysfunction relates to the severity of CHF [8] and affects central hemodynamics [9, 10]. It contributes both to reduced exercise capacity [11, 12] and the perception of dyspnea [2]. Furthermore, there is evidence that respiratory muscle dysfunction is an indicator of adverse prognosis in patients with CHF [13, 14]. Consequently, respiratory muscle has been the target of a substantial number of respiratory muscle training trials in patients with CHF [15–28]. Respiratory muscle training induces changes in measurements of respiratory muscle function such as PImax. These induced changes have been shown to correlate with enhanced exercise capacity, significantly less dyspnea, and improved NYHA functional status [19, 23–25, 27, 28].
123
Clin Res Cardiol
Despite the widespread application of measurements of respiratory muscle function, however, there are no data on biological variation, reference change value (RCV) or the minimal important difference (MID) in PImax in both healthy subjects and any group of patients, including those with CHF. The availability of these data would be useful in augmenting clinical decision making when changes in PImax occur that may or may not be clinically significant. We therefore determined the coefficients of variation, reference change values (RCVs), coefficients of reproducibility, and minimal important differences (MIDs) for PImax in patients with stable CHF at short-, intermediate-, and long-term follow-up intervals.
Methods
defined as evidence of cardiac decompensation, death from any cause, or cardiac transplantation in the 3 months following the second visit (V2)—the second PImax measurement was performed at V2; (c) no evidence of cardiac decompensation between V1 and V2; (d) no change between V1 and V2 in perceived symptoms or subjective exercise capacity as evidenced by the NYHA functional class. All patients were on stable CHF-specific medication for at least 1 month prior to inclusion in our study. Medication was at the discretion of the referring physician. Exclusion criteria were: a positive history for primary pulmonary disease; uncorrected valvular defects; cardiac decompensation requiring inotropic support within the last 3 months prior to study inclusion; and the presence of conditions that could possibly affect respiratory muscle function (e.g., thyroid dysfunction, electrolyte disturbance).
Patients Determination of clinical events and test intervals For the present analysis, we retrospectively selected all patients from the Heidelberg Clinical Heart Failure Registry of the Heart Failure Outpatients’ Clinic at the University of Heidelberg, Germany, meeting the criteria outlined below. The Heart Failure Outpatients’ Clinic is managed by specially trained cardiologists and nurses. The Heidelberg University Hospital fulfils both the role of a primary health care facility and that of a tertiary referral center for the Rhein-Neckar Region of Germany. Thus, patient cohorts selected represent a broad geographic and demographic range of patients. Patients visiting the Heart Failure Outpatients’ Clinic are asked to provide written, informed consent to the use of their data for research purposes and storage of these data in the Heidelberg Clinical Heart Failure Registry. Inclusion into this registry is continuous. The Heidelberg Clinical Heart Failure Registry itself and our protocol for the use of data from patients selected from this Registry met the principles outlined in the Declaration of Helsinki and were approved by the Ethics Committee of the University of Heidelberg. Inclusion/exclusion We included only patients with stable CHF due to systolic dysfunction. CHF was established according to published guidelines [29, 30] and systolic dysfunction was defined as a left-ventricular ejection fraction (LVEF) \45 %. Stability was assumed when all of the following criteria were met: (a) no evidence of cardiac decompensation in the 3 months preceding the inclusion visit (V1)—the initial PImax measurement of our study was performed at V1; (b) absence of any clinical event
123
The clinical events included in the definition of stability (see above) were cardiac decompensation, all-cause mortality, and cardiac transplantation. They were established from the patients’ clinical records, phone calls to the patient’s home or physician, or review of hospital in-patient records. Three groups of patients with different, distinct time intervals between PImax measurements were selected for our analyses: Group A had two measurements of PImax approximately 90 days apart, Group B had two measurements of PImax approximately 180 days apart, and Group C had two measurements of PImax approximately 365 days apart. Two measurements denote one measurement at V1 and one at V2. These intervals between measurements reflect short-term (90-day), intermediateterm (180-day), and long-term (365-day) biological variation from a clinical perspective because these intervals correspond either to the intervals commonly used to assess study effects or to the intervals commonly used for routine clinical follow-up. If a patient had more than 2 visits with measurement intervals that met the requirements for more than one group, inclusion into more than one group was allowed. This was assumed valid since the ‘‘learning effect’’ had been excluded already at inclusion into this study at V1 (see below) and all patients were stable between the two measurements of PImax irrespective of their group assignment. Overall, 302 measurements of PImax were performed in 260 different patients—thus 35 of the patients had one measurements of PImax counted towards two groups and 7 of the patients had one measurements of PImax counted towards all three groups.
Clin Res Cardiol
Determination of PImax To exclude a learning effect [31], at least one separate clinical visit to our outpatient department with determination of PImax prior to V1 of the present study was mandatory. Measurements of PImax were performed by a technician blinded to the status of the patient and the details of the study protocol. Patients took a deep breath through a flanged mouthpiece from the level of functional residual capacity against a shutter with a minor air leak preventing undesirable glottis closure (Erich Jaeger, MasterLabPro 4.2). Out of 3 measurements with \5 % variability between them, the highest pressure was recorded for data analysis. A short rest was allowed between maneuvres when necessary. PImax measurements are expressed in kPa units as positive values, despite being negative pressures with respect to atmospheric pressure. PImax was further expressed in percent of the individually predicted value following the equation by Harik-Khan et al. [32]. Statistics All analyses were performed separately for each group. In all groups, the total coefficient of variation (CVT) and the analytical coefficient of variation (CVA) for PImax values provided the basis for the determination of within-subject biological coefficient of variation (CVI) using the formula [33]: 1=2 CVI ¼ CV2T CV2A : Reference change values (RCVs) were calculated according to the formula: RCV ¼ z 21=2 ðCV2A =na þ CV2I =ns Þ1=2 ; where z = 1.96 (i.e., the z score for 95 % confidence with a two-tailed p \ 0.05); na is the number of replicate assays; and ns is the number of patient samples. The index of individuality (II) was calculated as follows: II = CVI =CVG ; where CVG is the between-subject biological CV for the group concerned [34]. The MID was determined using the one-standard error of measurement (one-SEM)-based approach developed by Wyrwich et al. [35, 36] following the equation: MID = SD SQRT ð1 rÞ; where SD is the population standard deviation, SQRT abbreviates square root, and r is the reliability coefficient (i.e., the degree of absolute agreement among measurements). We selected the intraclass correlation coefficient (ICC) as the reliability coefficient used in the above equation because it accounts for the proportion of variance
in test values due to between-subject variation; it is simple to calculate; and, it provides an estimate of MID that is in good agreement with other methods [35–37]. For the distribution-based approach to biological variation, RCV and MID represent mutually complementary concepts. While RCV relates closely to the idea of random variation around a homeostatic set point, MID addresses the idea of stability in the light of repeatability. To facilitate application, interpretation, and comparison of the RCV and MID values derived from our data, we transformed the individual values for RCV and MID into the corresponding relative or numerical values. Consequently the numerical MID values were transformed into relative values by normalization to the respective group mean while the relative values for RCV were transformed into numerical values via multiplication with the respective group median. The relationship between relative change in PImax and the respective PImax value at V1 for all cohorts was assessed following the method proposed by Bland and Altmann [38]. Descriptive statistics, the population SD, intraclass correlation coefficient, and paired samples t test values were obtained using MedCalc software version 12.7 (Ostend, Belgium) and results were displayed using GraphPad Prism version 6.02 for Windows (La Jolla, CA, USA). A p value of 0.05 was used to assess statistical significance.
Results Patient characteristics The clinical characteristics and exact follow-up intervals for all patients in Group A (90-day interval), Group B (180day interval), and Group C (365-day interval) are shown in Table 1. Non-ischemic heart disease was the underlying aetiology in the majority of patients and male gender was more prevalent in all groups. The mean ejection fraction was low with values between 22 and 31 % between groups, indicating more advanced heart failure for all three groups. The selection of stable groups was evidenced by the low rate of adverse clinical events after the second measurement of PImax in our study. At 1-year follow-up after V2, all-cause mortality was 4 % (n = 1) for Group A, 3 % (n = 3) for Group B, and 0.5 % (n = 1) for Group C. Cardiac decompensation at 1-year follow-up after V2 occurred in 4 % (n = 1) for Group A, 10 % (n = 9) for Group B, and 5 % (n = 9) for Group C patients. PImax and stability of measurements Respiratory muscle function was relatively preserved for all groups with respect to the achieved percentage of the
123
Clin Res Cardiol
individual predicted PImax [32]. Mean group percent values of individually predicted PImax were around 80 %; and 36–43 % of patients presented with measurements of PImax below 70 % of the individually predicted PImax (Table 1). For all three groups, the reliability coefficient, as expressed by the intraclass correlation coefficient between PImax measurements at V1 and V2 was above 0.8, indicating a high stability of measurements over time (Table 2). When analysed separately for the individual subgroups, values for the respective intraclass correlation coefficients were similar across all three groups with respect to age, gender, or aetiology of CHF. For complete results see Table 3.
Coefficient of variation and RCV
Summary statistics for PImax and MID
Comparison of RCV and MID
Summary statistics for PImax values—separate for V1 and V2 and each group—are shown in Table 2 along with the respective values for SD, r, and MID. The intraclass correlation coefficient for measurements of PImax at V1 and V2 was 0.83 in Group A, 0.87 in Group B, and 0.88 in Group C (Table 2). The corresponding values for MID ranged from 1.06 to 1.44 kPa among the three groups (Table 2).
Values for RCV among all groups were consistently lower than the normalized (relative) MID values of the respective group. Conversely, the numerical group values for MID were consistently higher than the transformed numerical RCV values for the respective group. Both relative and numerical differences, however, were very small indicating close approximation of values for RCV and MID.
Table 1 Patient demographics and PImax data at approximately 90-day (cohort A), 180-day (cohort B) and 365-day (cohort C) intervals
The distribution of measured values for PImax for each visit and group is shown in Fig. 1a–c, Bland–Altman plots, illustrating the % change in PImax values between V1 and V2 relative to the PImax value at V1 for all cohorts are shown in Fig. 2a–c. The respective values for the coefficients of variation, the resulting RCVs, and the IIs are presented in Table 4. Values for CVI and CVG ranged from 6.6 to 8.5 % and 37.1 to 42.7 %, respectively, for PImax values among patients in all three groups, while values for II ranged from 0.16 to 0.21. RCVs ranged from 10.8 to 13.6 % or 0.73 kPa to 0.97 kPa for Groups A to C.
Characteristic
Group A
Group B
Group C
N
25
93
184
Days between visits
93 (91–108)
185 (178–196)
365 (357–378)
No. of days of follow-upa
1365 (566–2459)
1405 (571–2819)
1593 (1066–2061)
Age (years)
54 ± 11
53 ± 10
57 ± 11
Sex (male)
20 (80)
78 (84)
148 (80)
27.6 ± 3.8
26.5 ± 5.0
27.7 ± 4.6
I
3 (12)
6 (7)
56 (30)
II
9 (36)
43 (46)
80 (46)
III
13 (52)
43 (46)
48 (28)
2
BMI (kg/m ) NYHA Class
0 (0)
0 (0)
0 (0.0)
LVEF (%)
IV
22 ± 8
25 ± 9
31 ± 12
ACE-I/ARB
24 (96)
87 (94)
180 (98)
Betablocker
13 (52)
59 (63)
167 (91)
IHD, n (%) PImax (kPa)
10 (40) 7.47 ± 3.53
44 (47) 7.53 ± 2.98
84 (46) 7.18 ± 3.13
PImax \70 % expectedb, n (%)
9 (36)
40 (43)
80 (43)
PImax/PImax expectedb, (%)
82 ± 38
82 ± 32
78 ± 31
All patients had stable CHF; values shown are mean ± SD or number (% of total) or median (IQR) were appropriate N number, M males, F females, BMI body mass index, NYHA New York Heart Association, LVEF leftventricular ejection fraction, IHD ischemic heart disease
123
a
After second 6WT measurement
b
According to formula by Harik-Khan et al. [32]
Clin Res Cardiol Table 2 Summary statistics for Pmax values (kPa) obtained in patients in cohorts A, B, and C at visits 1 (V1) and 2 (V2), *90 days (cohort A), *180 days (cohort B) and 365 days (cohort C) apart
Statistic
Group A (n = 25)
Group B (n = 93)
Group C (n = 184)
V1
V2
V1
V2
V1
V2
Mean
7.47
7.81
7.53
7.90
7.18
7.23
Change PIamax
0.35 ± 2.36
Median
7.16
7.90
6.76
7.90
6.72
6.79
SD
3.53
2.68
2.98
2.75
3.13
2.98
Range
1.66–17.72
2.01–12.64
1.21–15.35
2.49–15.85
1.13–16.43
1.13–16.94
r
0.8338
MID/mean (%)
19.3
MID
1.44
0.37 ± 1.90
0.06 ± 2.05
0.8730 18.4
0.8769
14.1
13.4
15.6
1.06
15.5
1.12
SD standard deviation, r intraclass correlation coefficient, MID minimum important difference a
Table 3 Intraclass correlation coefficients for subgroups defined by age, gender, or diagnosis
Mean change in PImax between V1 and V2
Characteristic
Intraclass correlation coefficient (95 % CI) Cohort A
Cohort B
Cohort C
0.83 (0.57 to 0.93) 0.80 (-0.24 to 0.98)
0.86 (0.78 to 0.91) 0.94 (0.83 to 0.98)
0.87 (0.81 to 0.90) 0.70 (0.41 to 0.85)
\Median
0.88 (0.44 to 0.97)
0.88 (0.78 to 0.94)
0.88 (0.81 to 0.92)
[Median
0.82 (0.44 to 0.94)
0.86 (0.76 to 0.92)
0.85 (0.78 to 0.90)
IHD
0.94 (0.51 to 0.99)
0.83 (0.64 to 0.92)
0.81 (0.66 to 0.89)
Other
0.81 (0.53 to 0.93)
0.89 (0.82 to 0.93)
0.89 (0.84 to 0.92)
Gender M F Age (years)
Diagnosis
CI confidence interval, M males, F females, IHD ischemic heart disease
A
90 days
B
20
180 days
365 days
C
20
5
V2
V1
5 0
0
0
10
V1
V2
V2
5
10
15
V1
10
15
Pimax (kPa)
15
Pimax (kPa)
Pimax (kPa)
20
Fig. 1 PImax values at visit 1 (V1) and visit 2 (V2) at approximately 90 days (a, cohort A), 180 days (b, cohort B) or 365 days (c, cohort C) between V1 and V2. Central box represents the limits between the 25th percentile (lower horizontal line of box) and 75th percentile
(upper horizontal line of box) of the data; line inside the central box represents the median of the data; lower and upper horizontal lines represent the range of the data
Discussion
induced by instability or intervention. In our study, measurements of PImax were highly stable for periods of up to 1 year. Following our results on biological variation, individually significant change in PImax values can be assumed beyond a change of 14 % (RCV) or 1.12 kPa (MID). To the best of our knowledge, this is the first study
Biological variation might be regarded as the random variation around a homeostatic set point of the respective test result [39]. It is inherent to any biological system and can only be measured in strict absence of any change
123
Clin Res Cardiol
A
B
90 days
0 -50
5
10
15
50 0 -50 -100
365 days 100
Relative Change (%)
50
-100
C
180 days 100
Relative Change (%)
Relative Change (%)
100
5
Pimax (kPa)
10
Pimax (kPa)
15
50 0 -50 -100
5
10
15
Pimax (kPa)
Fig. 2 Bland–Altman plot for PImax values at approximately 90 days (a, cohort A), 180 days (b, cohort B) or 365 days (c, cohort C). The dotted lines indicate 1.96 9 SD of the mean of the differences (95 % confidence with p \ 0.05)
Table 4 Analytical and biological variation of PImax at intervals of approximately 90 days (cohort A), 180 days (cohort B), and 365 days (cohort C) Group
PImax A (n = 25)
B (n = 93)
C (n = 184)
CVA (%)
2.0
2.0
2.0
CVaI (%)
8.5 (5.1–15.9)
7.2 (2.5–12.8)
6.6 (2.5–15.0)
CVaT (%)
8.7 (5.5–16.0)
7.5 (3.2–13.0)
6.9 (3.2–15.1)
CVG (%)
40.6
37.1
42.7
II (%)
21.4
20.2
16.3
RCV (%)
13.6
11.6
10.8
RCVb (kPa)
0.97
0.78
0.73
CV coefficient of variation, CVA analytical, CVI individual, CVG global, II index of individuality, RCV reference change values a
CVi and Cvt data are given as median (interquartile range)
b
RCV 9 median PImax of corresponding group
ever to address biological variation of PImax or to report on its proxies such as RCV or MID. Consequently, there is no other data against which we could directly compare our results. In general, there are two mutually complementary ways to approach the interpretation of time-dependent change in values for biological parameters such as PImax: anchorbased and distribution-based methods. The former requires reliable endpoints as ‘‘anchor’’ for definition while the latter requires stable cohorts for derivation of the respective descriptive (distribution) statistics of the cohort concerned. Our cohorts were selected for a maximum of stability— virtually eliminating clinical events in the first year of follow-up. We consider this a prerequisite for determination of biological variation. On the other hand, we would, therefore, be unable to reproduce and/or apply any anchorbased approach to our data. Our approach was distribution based. The most widely used proxies for assessing the potential clinical significance of time-dependent changes in values for biological parameters in distribution-based studies are RCV [39, 40] and
123
MID [41]. RCV mostly conveys the idea of statistical significance of change—addressing biological variation from the perspective of variation in the distribution of measurements. MID, on the other hand, mostly conveys the idea of clinical significance—addressing biological variation from the perspective of variation in the repeatability of measurements. To assess any potential differences between values for RCV and MID, we determined values for both. The similarity of RCV and MID values for PImax during all three of our observational periods underscores the robustness of our results. Furthermore, the low index of individuality suggests that conventional, population-based reference values are of limited utility for PImax. As a consequence, interpretation of a change in PImax should be based on RCV and/or MID values in conjunction with the results of other patient-specific findings. As indicated previously, there are no other data on biological variation of PImax to which we can directly compare our findings. In the setting of CHF we can, however, put our findings in perspective with the clinical training trials in CHF patients that report change in PImax as an interventional target [16, 18–21, 23, 27, 28]. On an individual patient data level, grouping of the respective control groups would enable evaluation of anchor-based strategies for biological variation. It is further beyond the scope of this article to discuss the possible relative contribution of specific training modalities to the magnitude of change in PImax values. It is, however, noteworthy that these training studies demonstrated changes from baseline in PImax ranging from 1.3 to 6.8 kPa (or 16–115 % in relative values) following completion of the respective training modality. There is only one trial that reported a change in PImax compared to baseline of only 0.44 kPa (6 %) in the non-intervention (aerobic exercise training only) group [20]. Clearly, these effects of respiratory muscle exercise training reported from the above trials exceed both our RCV and MID values. We suggest that knowledge of RCV and MID values for PImax contributes significantly to the
Clin Res Cardiol
interpretation of the respective trial results. It allows estimation of training effects beyond the mere demonstration of statistical significance between two median values of PImax. It is important to understand that any such change may not be clinically significant, whether it is statistically significant or not [42] as statistical significance between median values can be achieved even for small differences with sufficiently large a sample. It is here that demonstration of both RCV and MID values for PImax significantly enhances interpretation strategies. In fact, it would be our study that validates the results of the abovementioned trials as the change in PImax achieved would need to be interpreted against values for biological variation. In the light of our present results it is further possible that the only study ever to address the association of change in PImax with mortality [13] was inconclusive— despite the association of decreased PImax with increased mortality—because the observed mean change in PImax of 0.6 kPa (7.4 %) observed [13] was below the values for RCV and MID obtained in our present study. Another important aspect of our results is the high degree of stability of measurements of PImax in stable CHF patients for test intervals of 90 days up to 1 year. There is very little previously published data on this topic. One study reported both a high variability of measured values of PImax at first contact with the test and a strong initial learning effect [31]. The suitability and applicability of PImax as a measure of respiratory muscle function for clinical follow-up or as an endpoint target in any interventional trial could, therefore, be challenged. However, the findings from our study do not support such a challenge because the high coefficient of reproducibility (intraclass coefficient) of PImax observed in our study over a wide range of measurement intervals suggest strongly that values for PImax are useful in clinical trials on the beneficial effects of respiratory muscle exercise training in patients with CHF.
Limitations The results for biological variation, and especially MID, are context dependent. Our findings are therefore specific to the population of CHF patients included in our study. The number of patients included (especially in Group A) might be considered small. On the other hand, it compares well with the sample size in previously published studies on PImax in CHF. Also, we cannot completely rule out the presence of hidden clinical change, as we did not perform invasive testing or long-term Holter monitoring follow-up of our patients. Consequently, despite the rigorous definition of stability in our patients, it is conceivable that we did
not measure pure biological variation of PImax, but clinically undetected variation of CHF phenotype to a certain extent. However, this would argue in favour both of the stability of measurements of PImax and the RCV/MID values found in our study. This is because even with this potential confounder included, our stability of measurements of PImax is high and our RCV/MID values are very low. The high mean absolute level of PImax might represent a further limitation. It suggests relatively preserved respiratory muscle function. Therefore, the applicability of the RCV and MID values found in our study might be reduced for patients with very low PImax values. On the other hand, the range of PImax values found in our study (see Table 1) included patients with a more severely reduced PImax to a certain extent. This at least allows a sample-specific statement.
Conclusion In patients with stable chronic heart failure, measurements of PImax are highly stable for intervals up to 1 year. The low values for II suggest that evaluation of change in PImax should be performed on an individual (per patient) basis rather than following population-based cutoffs. Individually significant change in PImax values could be assumed beyond a change of 14 % (RCV) or 1.12 kPa (MID). Conflict of interest
None declared.
References 1. Hammond MD, Bauer KA, Sharp JT, Rocha RD (1990) Respiratory muscle strength in congestive heart failure. Chest 98(5):1091–1094 2. Mancini DM, Henson D, LaManca J, Levine S (1992) Respiratory muscle function and dyspnea in patients with chronic congestive heart failure. Circulation 86(3):909–918 3. Meyer FJ, Zugck C, Haass M, Otterspoor L, Strasser RH, Kubler W, Borst MM (2000) Inefficient ventilation and reduced respiratory muscle capacity in congestive heart failure. Basic Res Cardiol 95(4):333–342 4. Witt C, Borges AC, Haake H, Reindl I, Kleber FX, Baumann G (1997) Respiratory muscle weakness and normal ventilatory drive in dilative cardiomyopathy. Eur Heart J 18(8):1322–1328 5. Mancini DM, Henson D, LaManca J, Levine S (1994) Evidence of reduced respiratory muscle endurance in patients with heart failure. J Am Coll Cardiol 24(4):972–981 6. Walsh JT, Andrews R, Johnson P, Phillips L, Cowley AJ, Kinnear WJ (1996) Inspiratory muscle endurance in patients with chronic heart failure. Heart 76(4):332–336 7. Vincken W, Ghezzo H, Cosio MG (1987) Maximal static respiratory pressures in adults: normal values and their relationship to determinants of respiratory function. Bulletin Euro de PhysiopatholRes 23(5):435–439
123
Clin Res Cardiol 8. Ambrosino N, Opasich C, Crotti P, Cobelli F, Tavazzi L, Rampulla C (1994) Breathing pattern, ventilatory drive and respiratory muscle strength in patients with chronic heart failure. Euro Respir J Off J Euro Soc Clin Respir Physiol 7(1):17–22 9. Filusch A, Ewert R, Altesellmeier M, Zugck C, Hetzer R, Borst MM, Katus HA, Meyer FJ (2011) Respiratory muscle dysfunction in congestive heart failure–the role of pulmonary hypertension. Int J Cardiol 150(2):182–185. doi:10.1016/j.ijcard.2010.04.006 10. Nishimura Y, Maeda H, Tanaka K, Nakamura H, Hashimoto Y, Yokoyama M (1994) Respiratory muscle strength and hemodynamics in chronic heart failure. Chest 105(2):355–359 11. Ribeiro JP, Chiappa GR, Callegaro CC (2012) The contribution of inspiratory muscles function to exercise limitation in heart failure: pathophysiological mechanisms. Revista brasileira de fisioterapia 16(4):261–267 12. Ribeiro JP, Chiappa GR, Neder JA, Frankenstein L (2009) Respiratory muscle function and exercise intolerance in heart failure. Curr Heart Failure Reports 6(2):95–101 13. Frankenstein L, Nelles M, Meyer FJ, Sigg C, Schellberg D, Remppis BA, Katus HA, Zugck C (2009) Validity, prognostic value and optimal cutoff of respiratory muscle strength in patients with chronic heart failure changes with beta-blocker treatment. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups Epidemiol Prev Cardiac Rehabil Exerc Physiol 16(4):424–429. doi:10.1097/HJR.0b013e3283030a7e 14. Meyer FJ, Borst MM, Zugck C, Kirschke A, Schellberg D, Kubler W, Haass M (2001) Respiratory muscle dysfunction in congestive heart failure: clinical correlation and prognostic significance. Circulation 103(17):2153–2158 15. Chiappa GR, Roseguini BT, Vieira PJ, Alves CN, Tavares A, Winkelmann ER, Ferlin EL, Stein R, Ribeiro JP (2008) Inspiratory muscle training improves blood flow to resting and exercising limbs in patients with chronic heart failure. J Am Coll Cardiol 51(17):1663–1671. doi:10.1016/j.jacc.2007.12.045 16. Dall’Ago P, Chiappa GR, Guths H, Stein R, Ribeiro JP (2006) Inspiratory muscle training in patients with heart failure and inspiratory muscle weakness: a randomized trial. J Am Coll Cardiol 47(4):757–763. doi:10.1016/j.jacc.2005.09.052 17. Jaenisch RB, Hentschke VS, Quagliotto E, Cavinato PR, Schmeing LA, Xavier LL, Dal Lago P (2011) Respiratory muscle training improves hemodynamics, autonomic function, baroreceptor sensitivity, and respiratory mechanics in rats with heart failure. J Appl Physiol 111(6):1664–1670. doi:10.1152/japplphy siol.01245.2010 18. Johnson PH, Cowley AJ, Kinnear WJ (1998) A randomized controlled trial of inspiratory muscle training in stable chronic heart failure. Eur Heart J 19(8):1249–1253 19. Laoutaris I, Dritsas A, Brown MD, Manginas A, Alivizatos PA, Cokkinos DV (2004) Inspiratory muscle training using an incremental endurance test alleviates dyspnea and improves functional status in patients with chronic heart failure. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups Epidemiol Prev Cardiac Rehabil Exerc Physiol 11(6):489–496 20. Laoutaris ID, Adamopoulos S, Manginas A, Panagiotakos DB, Kallistratos MS, Doulaptsis C, Kouloubinis A, Voudris V, Pavlides G, Cokkinos DV, Dritsas A (2013) Benefits of combined aerobic/resistance/inspiratory training in patients with chronic heart failure. A complete exercise model? A prospective randomised study. Int J Cardiol 167(5):1967–1972. doi:10.1016/j. ijcard.2012.05.019 21. Laoutaris ID, Dritsas A, Adamopoulos S, Manginas A, Gouziouta A, Kallistratos MS, Koulopoulou M, Voudris V, Cokkinos DV, Sfirakis P (2011) Benefits of physical training on exercise capacity, inspiratory muscle function, and quality of life in patients with ventricular assist devices long-term postimplantation. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups
123
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
Epidemiol Prev Cardiac Rehabil Exerc Physiol 18(1):33–40. doi:10.1097/HJR.0b013e32833c0320 Laoutaris ID, Dritsas A, Brown MD, Manginas A, Kallistratos MS, Chaidaroglou A, Degiannis D, Alivizatos PA, Cokkinos DV (2008) Effects of inspiratory muscle training on autonomic activity, endothelial vasodilator function, and N-terminal pro-brain natriuretic peptide levels in chronic heart failure. J Cardiopulm Rehabil Prev 28(2):99–106. doi:10.1097/01.HCR.0000314203. 09676.b9 Mancini DM, Henson D, La Manca J, Donchez L, Levine S (1995) Benefit of selective respiratory muscle training on exercise capacity in patients with chronic congestive heart failure. Circulation 91(2):320–329 McConnell TR, Mandak JS, Sykes JS, Fesniak H, Dasgupta H (2003) Exercise training for heart failure patients improves respiratory muscle endurance, exercise tolerance, breathlessness, and quality of life. J Cardpulm Rehabil 23(1):10–16 Stein R, Chiappa GR, Guths H, Dall’Ago P, Ribeiro JP (2009) Inspiratory muscle training improves oxygen uptake efficiency slope in patients with chronic heart failure. J Cardiopulm Rehabil Prev 29(6):392–395. doi:10.1097/HCR.0b013e3181b4cc41 Tabet JY, Meurin P, Driss AB, Weber H, Renaud N, Grosdemouge A, Beauvais F, Cohen-Solal A (2009) Benefits of exercise training in chronic heart failure. Arch Cardiovasc Dis 102(10):721–730. doi:10.1016/j.acvd.2009.05.011 Weiner P, Waizman J, Magadle R, Berar-Yanay N, Pelled B (1999) The effect of specific inspiratory muscle training on the sensation of dyspnea and exercise tolerance in patients with congestive heart failure. Clin Cardiol 22(11):727–732 Winkelmann ER, Chiappa GR, Lima CO, Viecili PR, Stein R, Ribeiro JP (2009) Addition of inspiratory muscle training to aerobic training improves cardiorespiratory responses to exercise in patients with heart failure and inspiratory muscle weakness. Am Heart J 158 (5):768e761–768e767. doi:10.1016/j.ahj.2009.09.005 Hunt SA, American College of C, American Heart Association Task Force on Practice G (2005) ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 46(6):e1–e82. doi:10.1016/j.jacc.2005.08.022 McMurray JJ, Adamopoulos S, Anker SD, Auricchio A, Bohm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Gomez-Sanchez MA, Jaarsma T, Kober L, Lip GY, Maggioni AP, Parkhomenko A, Pieske BM, Popescu BA, Ronnevik PK, Rutten FH, Schwitter J, Seferovic P, Stepinska J, Trindade PT, Voors AA, Zannad F, Zeiher A, Guidelines ESCCfP (2012) ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012: The task force for the diagnosis and treatment of acute and chronic heart failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association (HFA) of the ESC. Euro Heart J 33(14):1787–1847. doi:10.1093/ eurheartj/ehs104 Wen AS, Woo MS, Keens TG (1997) How many maneuvers are required to measure maximal inspiratory pressure accurately. Chest 111(3):802–807 Harik-Khan RI, Wise RA, Fozard JL (1998) Determinants of maximal inspiratory pressure. The Baltimore longitudinal study of aging. Am J Respir Crit Care Med 158(5 Pt 1):1459–1464. doi:10.1164/ajrccm.158.5.9712006 Fraser CG, Harris EK (1989) Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 27(5):409–437. doi:10.3109/10408368909106595 Iglesias N, Petersen PH, Ricos C (2005) Power function of the reference change value in relation to cut-off points, reference
Clin Res Cardiol
35.
36.
37.
38.
intervals and index of individuality. Clin Chem Lab Med CCLM/ FESCC 43(4):441–448. doi:10.1515/CCLM.2005.078 Wyrwich KW, Nienaber NA, Tierney WM, Wolinsky FD (1999) Linking clinical relevance and statistical significance in evaluating intra-individual changes in health-related quality of life. Med Care 37(5):469–478 Wyrwich KW, Tierney WM, Wolinsky FD (1999) Further evidence supporting an SEM-based criterion for identifying meaningful intra-individual changes in health-related quality of life. J Clin Epidemiol 52(9):861–873 Wells G, Beaton D, Shea B, Boers M, Simon L, Strand V, Brooks P, Tugwell P (2001) Minimal clinically important differences: review of methods. J Rheumatol 28(2):406–412 Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1(8476):307–310
39. Fraser CG (2001) Changes in serial results. In: Fraser CG (ed) Biological variation: from principles to practice. AACC Press, Washington, pp 67–90 ´ lvarez V, Simo´n M, Biosca C, 40. Rico´s C, Perich C, Minchinela J, A Dome´nech M, Ferna´ndez P, Jime´nez CV, Garcia-Lario JV, Cava F (2009) Application of biological variation—a review. Biochem Med 19(3):250–259. doi:10.11613/BM.2009.023 41. Schunemann HJ, Guyatt GH (2005) Commentary–goodbye M(C)ID! Hello MID, where do you come from? Health Serv Res 40(2):593–597. doi:10.1111/j.1475-6773.2005.00374.x 42. Beaton DE, Boers M, Wells GA (2002) Many faces of the minimal clinically important difference (MCID): a literature review and directions for future research. Curr Opin Rheumatol 14(2):109–114
123
ORIGINAL PAPER
Biological variation, reference change value (RCV) and minimal important difference (MID) of inspiratory muscle strength (PImax) in patients with stable chronic heart failure Tobias Ta¨ger1 • Miriam Schell1 • Rita Cebola1 • Hanna Fro¨hlich1 • Andreas Do¨sch1 Jennifer Franke1 • Hugo A. Katus1 • Frank H. Wians Jr.2 • Lutz Frankenstein1
•
Received: 6 August 2014 / Accepted: 27 March 2015 Ó Springer-Verlag Berlin Heidelberg 2015
Abstract Background Despite the widespread application of measurements of respiratory muscle force (PImax) in clinical trials there is no data on biological variation, reference change value (RCV), or the minimal important difference (MID) for PImax irrespective of the target cohort. We addressed this issue for patients with chronic stable heart failure. Methods and results From the outpatients’ clinic of the University of Heidelberg we retrospectively selected three groups of patients with stable systolic chronic heart failure (CHF). Each group had two measurements of PImax: 90 days apart in Group A (n = 25), 180 days apart in Group B (n = 93), and 365 days apart in Group C (n = 184). Stability was defined as (a) no change in NYHA class between visits and (b) absence of cardiac decompensation 3 months prior, during, and 3 months after measurements. For each group, we determined within-subject (CVI), between-subject (CVG), and total (CVT) coefficient of variation (CV), the index of individuality (II), RCV, reliability coefficient, and MID of PImax. CVT was 8.7, 7.5, and 6.9 % for groups A, B, and C, respectively. The II and RCV were 0.21, 0.20, 0.16 and 13.6, 11.6, 10.8 %, respectively. The reliability coefficient and MID were 0.83, 0.87, 0.88 and 1.44, 1.06, 1.12 kPa, respectively. Results were similar between age, gender, and aetiology subgroups.
& Lutz Frankenstein [email protected] 1
Department of Cardiology, Angiology, and Pulmonology, University of Heidelberg, Im Neuenheimer Feld 410, 69120 Heidelberg, Germany
2
Department of Pathology, Baylor University Medical Center, Dallas, TX, USA
Conclusion In patients with stable CHF, measurements of PImax are highly stable for intervals up to 1 year. The low values for II suggest that evaluation of change in PImax should be performed on an individual (per patient) basis. Individually significant change can be assumed beyond 14 % (RCV) or 1.12 kPa (MID). Keywords Chronic heart failure Inspiratory muscle strength PImax Reference change value Minimal important difference Biological variation
Introduction Respiratory muscle dysfunction is an important co-morbid condition in patients with chronic heart failure (CHF). It presents both in the form of reduced respiratory muscle force [1–4] and reduced respiratory muscle endurance [5, 6]. Respiratory muscle force is estimated by measuring maximum inspiratory mouth occlusion pressure (PImax) [7]. In CHF patients, respiratory muscle dysfunction relates to the severity of CHF [8] and affects central hemodynamics [9, 10]. It contributes both to reduced exercise capacity [11, 12] and the perception of dyspnea [2]. Furthermore, there is evidence that respiratory muscle dysfunction is an indicator of adverse prognosis in patients with CHF [13, 14]. Consequently, respiratory muscle has been the target of a substantial number of respiratory muscle training trials in patients with CHF [15–28]. Respiratory muscle training induces changes in measurements of respiratory muscle function such as PImax. These induced changes have been shown to correlate with enhanced exercise capacity, significantly less dyspnea, and improved NYHA functional status [19, 23–25, 27, 28].
123
Clin Res Cardiol
Despite the widespread application of measurements of respiratory muscle function, however, there are no data on biological variation, reference change value (RCV) or the minimal important difference (MID) in PImax in both healthy subjects and any group of patients, including those with CHF. The availability of these data would be useful in augmenting clinical decision making when changes in PImax occur that may or may not be clinically significant. We therefore determined the coefficients of variation, reference change values (RCVs), coefficients of reproducibility, and minimal important differences (MIDs) for PImax in patients with stable CHF at short-, intermediate-, and long-term follow-up intervals.
Methods
defined as evidence of cardiac decompensation, death from any cause, or cardiac transplantation in the 3 months following the second visit (V2)—the second PImax measurement was performed at V2; (c) no evidence of cardiac decompensation between V1 and V2; (d) no change between V1 and V2 in perceived symptoms or subjective exercise capacity as evidenced by the NYHA functional class. All patients were on stable CHF-specific medication for at least 1 month prior to inclusion in our study. Medication was at the discretion of the referring physician. Exclusion criteria were: a positive history for primary pulmonary disease; uncorrected valvular defects; cardiac decompensation requiring inotropic support within the last 3 months prior to study inclusion; and the presence of conditions that could possibly affect respiratory muscle function (e.g., thyroid dysfunction, electrolyte disturbance).
Patients Determination of clinical events and test intervals For the present analysis, we retrospectively selected all patients from the Heidelberg Clinical Heart Failure Registry of the Heart Failure Outpatients’ Clinic at the University of Heidelberg, Germany, meeting the criteria outlined below. The Heart Failure Outpatients’ Clinic is managed by specially trained cardiologists and nurses. The Heidelberg University Hospital fulfils both the role of a primary health care facility and that of a tertiary referral center for the Rhein-Neckar Region of Germany. Thus, patient cohorts selected represent a broad geographic and demographic range of patients. Patients visiting the Heart Failure Outpatients’ Clinic are asked to provide written, informed consent to the use of their data for research purposes and storage of these data in the Heidelberg Clinical Heart Failure Registry. Inclusion into this registry is continuous. The Heidelberg Clinical Heart Failure Registry itself and our protocol for the use of data from patients selected from this Registry met the principles outlined in the Declaration of Helsinki and were approved by the Ethics Committee of the University of Heidelberg. Inclusion/exclusion We included only patients with stable CHF due to systolic dysfunction. CHF was established according to published guidelines [29, 30] and systolic dysfunction was defined as a left-ventricular ejection fraction (LVEF) \45 %. Stability was assumed when all of the following criteria were met: (a) no evidence of cardiac decompensation in the 3 months preceding the inclusion visit (V1)—the initial PImax measurement of our study was performed at V1; (b) absence of any clinical event
123
The clinical events included in the definition of stability (see above) were cardiac decompensation, all-cause mortality, and cardiac transplantation. They were established from the patients’ clinical records, phone calls to the patient’s home or physician, or review of hospital in-patient records. Three groups of patients with different, distinct time intervals between PImax measurements were selected for our analyses: Group A had two measurements of PImax approximately 90 days apart, Group B had two measurements of PImax approximately 180 days apart, and Group C had two measurements of PImax approximately 365 days apart. Two measurements denote one measurement at V1 and one at V2. These intervals between measurements reflect short-term (90-day), intermediateterm (180-day), and long-term (365-day) biological variation from a clinical perspective because these intervals correspond either to the intervals commonly used to assess study effects or to the intervals commonly used for routine clinical follow-up. If a patient had more than 2 visits with measurement intervals that met the requirements for more than one group, inclusion into more than one group was allowed. This was assumed valid since the ‘‘learning effect’’ had been excluded already at inclusion into this study at V1 (see below) and all patients were stable between the two measurements of PImax irrespective of their group assignment. Overall, 302 measurements of PImax were performed in 260 different patients—thus 35 of the patients had one measurements of PImax counted towards two groups and 7 of the patients had one measurements of PImax counted towards all three groups.
Clin Res Cardiol
Determination of PImax To exclude a learning effect [31], at least one separate clinical visit to our outpatient department with determination of PImax prior to V1 of the present study was mandatory. Measurements of PImax were performed by a technician blinded to the status of the patient and the details of the study protocol. Patients took a deep breath through a flanged mouthpiece from the level of functional residual capacity against a shutter with a minor air leak preventing undesirable glottis closure (Erich Jaeger, MasterLabPro 4.2). Out of 3 measurements with \5 % variability between them, the highest pressure was recorded for data analysis. A short rest was allowed between maneuvres when necessary. PImax measurements are expressed in kPa units as positive values, despite being negative pressures with respect to atmospheric pressure. PImax was further expressed in percent of the individually predicted value following the equation by Harik-Khan et al. [32]. Statistics All analyses were performed separately for each group. In all groups, the total coefficient of variation (CVT) and the analytical coefficient of variation (CVA) for PImax values provided the basis for the determination of within-subject biological coefficient of variation (CVI) using the formula [33]: 1=2 CVI ¼ CV2T CV2A : Reference change values (RCVs) were calculated according to the formula: RCV ¼ z 21=2 ðCV2A =na þ CV2I =ns Þ1=2 ; where z = 1.96 (i.e., the z score for 95 % confidence with a two-tailed p \ 0.05); na is the number of replicate assays; and ns is the number of patient samples. The index of individuality (II) was calculated as follows: II = CVI =CVG ; where CVG is the between-subject biological CV for the group concerned [34]. The MID was determined using the one-standard error of measurement (one-SEM)-based approach developed by Wyrwich et al. [35, 36] following the equation: MID = SD SQRT ð1 rÞ; where SD is the population standard deviation, SQRT abbreviates square root, and r is the reliability coefficient (i.e., the degree of absolute agreement among measurements). We selected the intraclass correlation coefficient (ICC) as the reliability coefficient used in the above equation because it accounts for the proportion of variance
in test values due to between-subject variation; it is simple to calculate; and, it provides an estimate of MID that is in good agreement with other methods [35–37]. For the distribution-based approach to biological variation, RCV and MID represent mutually complementary concepts. While RCV relates closely to the idea of random variation around a homeostatic set point, MID addresses the idea of stability in the light of repeatability. To facilitate application, interpretation, and comparison of the RCV and MID values derived from our data, we transformed the individual values for RCV and MID into the corresponding relative or numerical values. Consequently the numerical MID values were transformed into relative values by normalization to the respective group mean while the relative values for RCV were transformed into numerical values via multiplication with the respective group median. The relationship between relative change in PImax and the respective PImax value at V1 for all cohorts was assessed following the method proposed by Bland and Altmann [38]. Descriptive statistics, the population SD, intraclass correlation coefficient, and paired samples t test values were obtained using MedCalc software version 12.7 (Ostend, Belgium) and results were displayed using GraphPad Prism version 6.02 for Windows (La Jolla, CA, USA). A p value of 0.05 was used to assess statistical significance.
Results Patient characteristics The clinical characteristics and exact follow-up intervals for all patients in Group A (90-day interval), Group B (180day interval), and Group C (365-day interval) are shown in Table 1. Non-ischemic heart disease was the underlying aetiology in the majority of patients and male gender was more prevalent in all groups. The mean ejection fraction was low with values between 22 and 31 % between groups, indicating more advanced heart failure for all three groups. The selection of stable groups was evidenced by the low rate of adverse clinical events after the second measurement of PImax in our study. At 1-year follow-up after V2, all-cause mortality was 4 % (n = 1) for Group A, 3 % (n = 3) for Group B, and 0.5 % (n = 1) for Group C. Cardiac decompensation at 1-year follow-up after V2 occurred in 4 % (n = 1) for Group A, 10 % (n = 9) for Group B, and 5 % (n = 9) for Group C patients. PImax and stability of measurements Respiratory muscle function was relatively preserved for all groups with respect to the achieved percentage of the
123
Clin Res Cardiol
individual predicted PImax [32]. Mean group percent values of individually predicted PImax were around 80 %; and 36–43 % of patients presented with measurements of PImax below 70 % of the individually predicted PImax (Table 1). For all three groups, the reliability coefficient, as expressed by the intraclass correlation coefficient between PImax measurements at V1 and V2 was above 0.8, indicating a high stability of measurements over time (Table 2). When analysed separately for the individual subgroups, values for the respective intraclass correlation coefficients were similar across all three groups with respect to age, gender, or aetiology of CHF. For complete results see Table 3.
Coefficient of variation and RCV
Summary statistics for PImax and MID
Comparison of RCV and MID
Summary statistics for PImax values—separate for V1 and V2 and each group—are shown in Table 2 along with the respective values for SD, r, and MID. The intraclass correlation coefficient for measurements of PImax at V1 and V2 was 0.83 in Group A, 0.87 in Group B, and 0.88 in Group C (Table 2). The corresponding values for MID ranged from 1.06 to 1.44 kPa among the three groups (Table 2).
Values for RCV among all groups were consistently lower than the normalized (relative) MID values of the respective group. Conversely, the numerical group values for MID were consistently higher than the transformed numerical RCV values for the respective group. Both relative and numerical differences, however, were very small indicating close approximation of values for RCV and MID.
Table 1 Patient demographics and PImax data at approximately 90-day (cohort A), 180-day (cohort B) and 365-day (cohort C) intervals
The distribution of measured values for PImax for each visit and group is shown in Fig. 1a–c, Bland–Altman plots, illustrating the % change in PImax values between V1 and V2 relative to the PImax value at V1 for all cohorts are shown in Fig. 2a–c. The respective values for the coefficients of variation, the resulting RCVs, and the IIs are presented in Table 4. Values for CVI and CVG ranged from 6.6 to 8.5 % and 37.1 to 42.7 %, respectively, for PImax values among patients in all three groups, while values for II ranged from 0.16 to 0.21. RCVs ranged from 10.8 to 13.6 % or 0.73 kPa to 0.97 kPa for Groups A to C.
Characteristic
Group A
Group B
Group C
N
25
93
184
Days between visits
93 (91–108)
185 (178–196)
365 (357–378)
No. of days of follow-upa
1365 (566–2459)
1405 (571–2819)
1593 (1066–2061)
Age (years)
54 ± 11
53 ± 10
57 ± 11
Sex (male)
20 (80)
78 (84)
148 (80)
27.6 ± 3.8
26.5 ± 5.0
27.7 ± 4.6
I
3 (12)
6 (7)
56 (30)
II
9 (36)
43 (46)
80 (46)
III
13 (52)
43 (46)
48 (28)
2
BMI (kg/m ) NYHA Class
0 (0)
0 (0)
0 (0.0)
LVEF (%)
IV
22 ± 8
25 ± 9
31 ± 12
ACE-I/ARB
24 (96)
87 (94)
180 (98)
Betablocker
13 (52)
59 (63)
167 (91)
IHD, n (%) PImax (kPa)
10 (40) 7.47 ± 3.53
44 (47) 7.53 ± 2.98
84 (46) 7.18 ± 3.13
PImax \70 % expectedb, n (%)
9 (36)
40 (43)
80 (43)
PImax/PImax expectedb, (%)
82 ± 38
82 ± 32
78 ± 31
All patients had stable CHF; values shown are mean ± SD or number (% of total) or median (IQR) were appropriate N number, M males, F females, BMI body mass index, NYHA New York Heart Association, LVEF leftventricular ejection fraction, IHD ischemic heart disease
123
a
After second 6WT measurement
b
According to formula by Harik-Khan et al. [32]
Clin Res Cardiol Table 2 Summary statistics for Pmax values (kPa) obtained in patients in cohorts A, B, and C at visits 1 (V1) and 2 (V2), *90 days (cohort A), *180 days (cohort B) and 365 days (cohort C) apart
Statistic
Group A (n = 25)
Group B (n = 93)
Group C (n = 184)
V1
V2
V1
V2
V1
V2
Mean
7.47
7.81
7.53
7.90
7.18
7.23
Change PIamax
0.35 ± 2.36
Median
7.16
7.90
6.76
7.90
6.72
6.79
SD
3.53
2.68
2.98
2.75
3.13
2.98
Range
1.66–17.72
2.01–12.64
1.21–15.35
2.49–15.85
1.13–16.43
1.13–16.94
r
0.8338
MID/mean (%)
19.3
MID
1.44
0.37 ± 1.90
0.06 ± 2.05
0.8730 18.4
0.8769
14.1
13.4
15.6
1.06
15.5
1.12
SD standard deviation, r intraclass correlation coefficient, MID minimum important difference a
Table 3 Intraclass correlation coefficients for subgroups defined by age, gender, or diagnosis
Mean change in PImax between V1 and V2
Characteristic
Intraclass correlation coefficient (95 % CI) Cohort A
Cohort B
Cohort C
0.83 (0.57 to 0.93) 0.80 (-0.24 to 0.98)
0.86 (0.78 to 0.91) 0.94 (0.83 to 0.98)
0.87 (0.81 to 0.90) 0.70 (0.41 to 0.85)
\Median
0.88 (0.44 to 0.97)
0.88 (0.78 to 0.94)
0.88 (0.81 to 0.92)
[Median
0.82 (0.44 to 0.94)
0.86 (0.76 to 0.92)
0.85 (0.78 to 0.90)
IHD
0.94 (0.51 to 0.99)
0.83 (0.64 to 0.92)
0.81 (0.66 to 0.89)
Other
0.81 (0.53 to 0.93)
0.89 (0.82 to 0.93)
0.89 (0.84 to 0.92)
Gender M F Age (years)
Diagnosis
CI confidence interval, M males, F females, IHD ischemic heart disease
A
90 days
B
20
180 days
365 days
C
20
5
V2
V1
5 0
0
0
10
V1
V2
V2
5
10
15
V1
10
15
Pimax (kPa)
15
Pimax (kPa)
Pimax (kPa)
20
Fig. 1 PImax values at visit 1 (V1) and visit 2 (V2) at approximately 90 days (a, cohort A), 180 days (b, cohort B) or 365 days (c, cohort C) between V1 and V2. Central box represents the limits between the 25th percentile (lower horizontal line of box) and 75th percentile
(upper horizontal line of box) of the data; line inside the central box represents the median of the data; lower and upper horizontal lines represent the range of the data
Discussion
induced by instability or intervention. In our study, measurements of PImax were highly stable for periods of up to 1 year. Following our results on biological variation, individually significant change in PImax values can be assumed beyond a change of 14 % (RCV) or 1.12 kPa (MID). To the best of our knowledge, this is the first study
Biological variation might be regarded as the random variation around a homeostatic set point of the respective test result [39]. It is inherent to any biological system and can only be measured in strict absence of any change
123
Clin Res Cardiol
A
B
90 days
0 -50
5
10
15
50 0 -50 -100
365 days 100
Relative Change (%)
50
-100
C
180 days 100
Relative Change (%)
Relative Change (%)
100
5
Pimax (kPa)
10
Pimax (kPa)
15
50 0 -50 -100
5
10
15
Pimax (kPa)
Fig. 2 Bland–Altman plot for PImax values at approximately 90 days (a, cohort A), 180 days (b, cohort B) or 365 days (c, cohort C). The dotted lines indicate 1.96 9 SD of the mean of the differences (95 % confidence with p \ 0.05)
Table 4 Analytical and biological variation of PImax at intervals of approximately 90 days (cohort A), 180 days (cohort B), and 365 days (cohort C) Group
PImax A (n = 25)
B (n = 93)
C (n = 184)
CVA (%)
2.0
2.0
2.0
CVaI (%)
8.5 (5.1–15.9)
7.2 (2.5–12.8)
6.6 (2.5–15.0)
CVaT (%)
8.7 (5.5–16.0)
7.5 (3.2–13.0)
6.9 (3.2–15.1)
CVG (%)
40.6
37.1
42.7
II (%)
21.4
20.2
16.3
RCV (%)
13.6
11.6
10.8
RCVb (kPa)
0.97
0.78
0.73
CV coefficient of variation, CVA analytical, CVI individual, CVG global, II index of individuality, RCV reference change values a
CVi and Cvt data are given as median (interquartile range)
b
RCV 9 median PImax of corresponding group
ever to address biological variation of PImax or to report on its proxies such as RCV or MID. Consequently, there is no other data against which we could directly compare our results. In general, there are two mutually complementary ways to approach the interpretation of time-dependent change in values for biological parameters such as PImax: anchorbased and distribution-based methods. The former requires reliable endpoints as ‘‘anchor’’ for definition while the latter requires stable cohorts for derivation of the respective descriptive (distribution) statistics of the cohort concerned. Our cohorts were selected for a maximum of stability— virtually eliminating clinical events in the first year of follow-up. We consider this a prerequisite for determination of biological variation. On the other hand, we would, therefore, be unable to reproduce and/or apply any anchorbased approach to our data. Our approach was distribution based. The most widely used proxies for assessing the potential clinical significance of time-dependent changes in values for biological parameters in distribution-based studies are RCV [39, 40] and
123
MID [41]. RCV mostly conveys the idea of statistical significance of change—addressing biological variation from the perspective of variation in the distribution of measurements. MID, on the other hand, mostly conveys the idea of clinical significance—addressing biological variation from the perspective of variation in the repeatability of measurements. To assess any potential differences between values for RCV and MID, we determined values for both. The similarity of RCV and MID values for PImax during all three of our observational periods underscores the robustness of our results. Furthermore, the low index of individuality suggests that conventional, population-based reference values are of limited utility for PImax. As a consequence, interpretation of a change in PImax should be based on RCV and/or MID values in conjunction with the results of other patient-specific findings. As indicated previously, there are no other data on biological variation of PImax to which we can directly compare our findings. In the setting of CHF we can, however, put our findings in perspective with the clinical training trials in CHF patients that report change in PImax as an interventional target [16, 18–21, 23, 27, 28]. On an individual patient data level, grouping of the respective control groups would enable evaluation of anchor-based strategies for biological variation. It is further beyond the scope of this article to discuss the possible relative contribution of specific training modalities to the magnitude of change in PImax values. It is, however, noteworthy that these training studies demonstrated changes from baseline in PImax ranging from 1.3 to 6.8 kPa (or 16–115 % in relative values) following completion of the respective training modality. There is only one trial that reported a change in PImax compared to baseline of only 0.44 kPa (6 %) in the non-intervention (aerobic exercise training only) group [20]. Clearly, these effects of respiratory muscle exercise training reported from the above trials exceed both our RCV and MID values. We suggest that knowledge of RCV and MID values for PImax contributes significantly to the
Clin Res Cardiol
interpretation of the respective trial results. It allows estimation of training effects beyond the mere demonstration of statistical significance between two median values of PImax. It is important to understand that any such change may not be clinically significant, whether it is statistically significant or not [42] as statistical significance between median values can be achieved even for small differences with sufficiently large a sample. It is here that demonstration of both RCV and MID values for PImax significantly enhances interpretation strategies. In fact, it would be our study that validates the results of the abovementioned trials as the change in PImax achieved would need to be interpreted against values for biological variation. In the light of our present results it is further possible that the only study ever to address the association of change in PImax with mortality [13] was inconclusive— despite the association of decreased PImax with increased mortality—because the observed mean change in PImax of 0.6 kPa (7.4 %) observed [13] was below the values for RCV and MID obtained in our present study. Another important aspect of our results is the high degree of stability of measurements of PImax in stable CHF patients for test intervals of 90 days up to 1 year. There is very little previously published data on this topic. One study reported both a high variability of measured values of PImax at first contact with the test and a strong initial learning effect [31]. The suitability and applicability of PImax as a measure of respiratory muscle function for clinical follow-up or as an endpoint target in any interventional trial could, therefore, be challenged. However, the findings from our study do not support such a challenge because the high coefficient of reproducibility (intraclass coefficient) of PImax observed in our study over a wide range of measurement intervals suggest strongly that values for PImax are useful in clinical trials on the beneficial effects of respiratory muscle exercise training in patients with CHF.
Limitations The results for biological variation, and especially MID, are context dependent. Our findings are therefore specific to the population of CHF patients included in our study. The number of patients included (especially in Group A) might be considered small. On the other hand, it compares well with the sample size in previously published studies on PImax in CHF. Also, we cannot completely rule out the presence of hidden clinical change, as we did not perform invasive testing or long-term Holter monitoring follow-up of our patients. Consequently, despite the rigorous definition of stability in our patients, it is conceivable that we did
not measure pure biological variation of PImax, but clinically undetected variation of CHF phenotype to a certain extent. However, this would argue in favour both of the stability of measurements of PImax and the RCV/MID values found in our study. This is because even with this potential confounder included, our stability of measurements of PImax is high and our RCV/MID values are very low. The high mean absolute level of PImax might represent a further limitation. It suggests relatively preserved respiratory muscle function. Therefore, the applicability of the RCV and MID values found in our study might be reduced for patients with very low PImax values. On the other hand, the range of PImax values found in our study (see Table 1) included patients with a more severely reduced PImax to a certain extent. This at least allows a sample-specific statement.
Conclusion In patients with stable chronic heart failure, measurements of PImax are highly stable for intervals up to 1 year. The low values for II suggest that evaluation of change in PImax should be performed on an individual (per patient) basis rather than following population-based cutoffs. Individually significant change in PImax values could be assumed beyond a change of 14 % (RCV) or 1.12 kPa (MID). Conflict of interest
None declared.
References 1. Hammond MD, Bauer KA, Sharp JT, Rocha RD (1990) Respiratory muscle strength in congestive heart failure. Chest 98(5):1091–1094 2. Mancini DM, Henson D, LaManca J, Levine S (1992) Respiratory muscle function and dyspnea in patients with chronic congestive heart failure. Circulation 86(3):909–918 3. Meyer FJ, Zugck C, Haass M, Otterspoor L, Strasser RH, Kubler W, Borst MM (2000) Inefficient ventilation and reduced respiratory muscle capacity in congestive heart failure. Basic Res Cardiol 95(4):333–342 4. Witt C, Borges AC, Haake H, Reindl I, Kleber FX, Baumann G (1997) Respiratory muscle weakness and normal ventilatory drive in dilative cardiomyopathy. Eur Heart J 18(8):1322–1328 5. Mancini DM, Henson D, LaManca J, Levine S (1994) Evidence of reduced respiratory muscle endurance in patients with heart failure. J Am Coll Cardiol 24(4):972–981 6. Walsh JT, Andrews R, Johnson P, Phillips L, Cowley AJ, Kinnear WJ (1996) Inspiratory muscle endurance in patients with chronic heart failure. Heart 76(4):332–336 7. Vincken W, Ghezzo H, Cosio MG (1987) Maximal static respiratory pressures in adults: normal values and their relationship to determinants of respiratory function. Bulletin Euro de PhysiopatholRes 23(5):435–439
123
Clin Res Cardiol 8. Ambrosino N, Opasich C, Crotti P, Cobelli F, Tavazzi L, Rampulla C (1994) Breathing pattern, ventilatory drive and respiratory muscle strength in patients with chronic heart failure. Euro Respir J Off J Euro Soc Clin Respir Physiol 7(1):17–22 9. Filusch A, Ewert R, Altesellmeier M, Zugck C, Hetzer R, Borst MM, Katus HA, Meyer FJ (2011) Respiratory muscle dysfunction in congestive heart failure–the role of pulmonary hypertension. Int J Cardiol 150(2):182–185. doi:10.1016/j.ijcard.2010.04.006 10. Nishimura Y, Maeda H, Tanaka K, Nakamura H, Hashimoto Y, Yokoyama M (1994) Respiratory muscle strength and hemodynamics in chronic heart failure. Chest 105(2):355–359 11. Ribeiro JP, Chiappa GR, Callegaro CC (2012) The contribution of inspiratory muscles function to exercise limitation in heart failure: pathophysiological mechanisms. Revista brasileira de fisioterapia 16(4):261–267 12. Ribeiro JP, Chiappa GR, Neder JA, Frankenstein L (2009) Respiratory muscle function and exercise intolerance in heart failure. Curr Heart Failure Reports 6(2):95–101 13. Frankenstein L, Nelles M, Meyer FJ, Sigg C, Schellberg D, Remppis BA, Katus HA, Zugck C (2009) Validity, prognostic value and optimal cutoff of respiratory muscle strength in patients with chronic heart failure changes with beta-blocker treatment. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups Epidemiol Prev Cardiac Rehabil Exerc Physiol 16(4):424–429. doi:10.1097/HJR.0b013e3283030a7e 14. Meyer FJ, Borst MM, Zugck C, Kirschke A, Schellberg D, Kubler W, Haass M (2001) Respiratory muscle dysfunction in congestive heart failure: clinical correlation and prognostic significance. Circulation 103(17):2153–2158 15. Chiappa GR, Roseguini BT, Vieira PJ, Alves CN, Tavares A, Winkelmann ER, Ferlin EL, Stein R, Ribeiro JP (2008) Inspiratory muscle training improves blood flow to resting and exercising limbs in patients with chronic heart failure. J Am Coll Cardiol 51(17):1663–1671. doi:10.1016/j.jacc.2007.12.045 16. Dall’Ago P, Chiappa GR, Guths H, Stein R, Ribeiro JP (2006) Inspiratory muscle training in patients with heart failure and inspiratory muscle weakness: a randomized trial. J Am Coll Cardiol 47(4):757–763. doi:10.1016/j.jacc.2005.09.052 17. Jaenisch RB, Hentschke VS, Quagliotto E, Cavinato PR, Schmeing LA, Xavier LL, Dal Lago P (2011) Respiratory muscle training improves hemodynamics, autonomic function, baroreceptor sensitivity, and respiratory mechanics in rats with heart failure. J Appl Physiol 111(6):1664–1670. doi:10.1152/japplphy siol.01245.2010 18. Johnson PH, Cowley AJ, Kinnear WJ (1998) A randomized controlled trial of inspiratory muscle training in stable chronic heart failure. Eur Heart J 19(8):1249–1253 19. Laoutaris I, Dritsas A, Brown MD, Manginas A, Alivizatos PA, Cokkinos DV (2004) Inspiratory muscle training using an incremental endurance test alleviates dyspnea and improves functional status in patients with chronic heart failure. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups Epidemiol Prev Cardiac Rehabil Exerc Physiol 11(6):489–496 20. Laoutaris ID, Adamopoulos S, Manginas A, Panagiotakos DB, Kallistratos MS, Doulaptsis C, Kouloubinis A, Voudris V, Pavlides G, Cokkinos DV, Dritsas A (2013) Benefits of combined aerobic/resistance/inspiratory training in patients with chronic heart failure. A complete exercise model? A prospective randomised study. Int J Cardiol 167(5):1967–1972. doi:10.1016/j. ijcard.2012.05.019 21. Laoutaris ID, Dritsas A, Adamopoulos S, Manginas A, Gouziouta A, Kallistratos MS, Koulopoulou M, Voudris V, Cokkinos DV, Sfirakis P (2011) Benefits of physical training on exercise capacity, inspiratory muscle function, and quality of life in patients with ventricular assist devices long-term postimplantation. Euro J Cardiovasc Prev Rehabil Off J Euro Soc Cardiol Working Groups
123
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
Epidemiol Prev Cardiac Rehabil Exerc Physiol 18(1):33–40. doi:10.1097/HJR.0b013e32833c0320 Laoutaris ID, Dritsas A, Brown MD, Manginas A, Kallistratos MS, Chaidaroglou A, Degiannis D, Alivizatos PA, Cokkinos DV (2008) Effects of inspiratory muscle training on autonomic activity, endothelial vasodilator function, and N-terminal pro-brain natriuretic peptide levels in chronic heart failure. J Cardiopulm Rehabil Prev 28(2):99–106. doi:10.1097/01.HCR.0000314203. 09676.b9 Mancini DM, Henson D, La Manca J, Donchez L, Levine S (1995) Benefit of selective respiratory muscle training on exercise capacity in patients with chronic congestive heart failure. Circulation 91(2):320–329 McConnell TR, Mandak JS, Sykes JS, Fesniak H, Dasgupta H (2003) Exercise training for heart failure patients improves respiratory muscle endurance, exercise tolerance, breathlessness, and quality of life. J Cardpulm Rehabil 23(1):10–16 Stein R, Chiappa GR, Guths H, Dall’Ago P, Ribeiro JP (2009) Inspiratory muscle training improves oxygen uptake efficiency slope in patients with chronic heart failure. J Cardiopulm Rehabil Prev 29(6):392–395. doi:10.1097/HCR.0b013e3181b4cc41 Tabet JY, Meurin P, Driss AB, Weber H, Renaud N, Grosdemouge A, Beauvais F, Cohen-Solal A (2009) Benefits of exercise training in chronic heart failure. Arch Cardiovasc Dis 102(10):721–730. doi:10.1016/j.acvd.2009.05.011 Weiner P, Waizman J, Magadle R, Berar-Yanay N, Pelled B (1999) The effect of specific inspiratory muscle training on the sensation of dyspnea and exercise tolerance in patients with congestive heart failure. Clin Cardiol 22(11):727–732 Winkelmann ER, Chiappa GR, Lima CO, Viecili PR, Stein R, Ribeiro JP (2009) Addition of inspiratory muscle training to aerobic training improves cardiorespiratory responses to exercise in patients with heart failure and inspiratory muscle weakness. Am Heart J 158 (5):768e761–768e767. doi:10.1016/j.ahj.2009.09.005 Hunt SA, American College of C, American Heart Association Task Force on Practice G (2005) ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology/ American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 46(6):e1–e82. doi:10.1016/j.jacc.2005.08.022 McMurray JJ, Adamopoulos S, Anker SD, Auricchio A, Bohm M, Dickstein K, Falk V, Filippatos G, Fonseca C, Gomez-Sanchez MA, Jaarsma T, Kober L, Lip GY, Maggioni AP, Parkhomenko A, Pieske BM, Popescu BA, Ronnevik PK, Rutten FH, Schwitter J, Seferovic P, Stepinska J, Trindade PT, Voors AA, Zannad F, Zeiher A, Guidelines ESCCfP (2012) ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2012: The task force for the diagnosis and treatment of acute and chronic heart failure 2012 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association (HFA) of the ESC. Euro Heart J 33(14):1787–1847. doi:10.1093/ eurheartj/ehs104 Wen AS, Woo MS, Keens TG (1997) How many maneuvers are required to measure maximal inspiratory pressure accurately. Chest 111(3):802–807 Harik-Khan RI, Wise RA, Fozard JL (1998) Determinants of maximal inspiratory pressure. The Baltimore longitudinal study of aging. Am J Respir Crit Care Med 158(5 Pt 1):1459–1464. doi:10.1164/ajrccm.158.5.9712006 Fraser CG, Harris EK (1989) Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 27(5):409–437. doi:10.3109/10408368909106595 Iglesias N, Petersen PH, Ricos C (2005) Power function of the reference change value in relation to cut-off points, reference
Clin Res Cardiol
35.
36.
37.
38.
intervals and index of individuality. Clin Chem Lab Med CCLM/ FESCC 43(4):441–448. doi:10.1515/CCLM.2005.078 Wyrwich KW, Nienaber NA, Tierney WM, Wolinsky FD (1999) Linking clinical relevance and statistical significance in evaluating intra-individual changes in health-related quality of life. Med Care 37(5):469–478 Wyrwich KW, Tierney WM, Wolinsky FD (1999) Further evidence supporting an SEM-based criterion for identifying meaningful intra-individual changes in health-related quality of life. J Clin Epidemiol 52(9):861–873 Wells G, Beaton D, Shea B, Boers M, Simon L, Strand V, Brooks P, Tugwell P (2001) Minimal clinically important differences: review of methods. J Rheumatol 28(2):406–412 Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1(8476):307–310
39. Fraser CG (2001) Changes in serial results. In: Fraser CG (ed) Biological variation: from principles to practice. AACC Press, Washington, pp 67–90 ´ lvarez V, Simo´n M, Biosca C, 40. Rico´s C, Perich C, Minchinela J, A Dome´nech M, Ferna´ndez P, Jime´nez CV, Garcia-Lario JV, Cava F (2009) Application of biological variation—a review. Biochem Med 19(3):250–259. doi:10.11613/BM.2009.023 41. Schunemann HJ, Guyatt GH (2005) Commentary–goodbye M(C)ID! Hello MID, where do you come from? Health Serv Res 40(2):593–597. doi:10.1111/j.1475-6773.2005.00374.x 42. Beaton DE, Boers M, Wells GA (2002) Many faces of the minimal clinically important difference (MCID): a literature review and directions for future research. Curr Opin Rheumatol 14(2):109–114
123
