INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: spirometry; telemedicine; transplantation

Pulmonary function in lung transplant recipients is sensitive to infection and rejection, which often produce an obstructive ventilatory defect (1). Previous studies using bronchoalveolar lavage (BAL) and transbronchial lung biopsy (TBB) as gold standards have reported that the sensitivity of FEV₁ for the detection of these complications ranges from 60% to 75% in recipients of heart-lung (HLT) and bilateral-lung (BLT) transplantation (1, 3) and from 48% to 72% in recipients of single-lung transplantation (4). These observations have led to the widely accepted recommendation that the follow-up of lung transplant recipients should include daily measurements of FEV₁ at home with a portable spirometer (5, 6). The reliability of such monitoring, however, may be limited by several factors. First, the collaboration and understanding of the patient are mandatory as the patient has to analyze the functional results and to contact the transplant center when a significant change occurs. Second, because the transplant team has no direct access to the data files, these are not readily available for daily review. Finally, the portable spirometers that have been used so far (5, 6) provided values for FEV₁ and FVC but not for the mid-expiratory flow rate (FEF_25-75), although this variable seems to be more sensitive than FEV₁ for the detection of acute allograft dysfunction (3).

To overcome these potential limitations, one American (7) and two European (10, 11) centers have developed systems that allow daily or weekly telemetric transmission of home spirometric data to the hospital. These groups have reported very good adherence of transplanted patients to the program and excellent agreement between home and hospital spirometry (7), and they have shown that home spirometric measurements are reliable and valid (9). In line with these studies, we have recently developed and established the feasibility of a fully automated, user-friendly Internet-based infrastructure to monitor pulmonary function at home after transplantation (12). The primary aim of the present study was to determine the sensitivity and positive predictive value (PPV) of such monitoring for the detection of acute complications affecting the allograft.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Twenty-two HLT (n = 7) and BLT (n = 15) recipients were studied between June 1998 and September 2000 (Table 1); 19 patients were included in the study within days after discharge from hospital and three patients were included 97-222 d after surgery. We used an Internet-based infrastructure consisting of a PC-based user-friendly terminal with a keypad and a screen, a modem, and a microspirometer (Microloop II; Micro Medical, Rochester, UK) at home, and a central server with a database at the hospital (12). Patients were instructed to perform one measurement session in the morning and one in the evening. A session consisted of answering a symptom questionnaire (temperature, breathlessness, sputum, chest pain, fatigue) and performing one to three forced expiratory maneuvers. The computer requested up to two additional maneuvers whenever the first FEV₁ value was below 100% of the reference value, which was regularly updated as lung function improved after surgery. The flow-volume loop with the highest FEV₁ was transmitted to the transplant center. Hospital spirometry was obtained with a Sensormedics 2400 unit (Sensormedics, Anaheim, CA).

Data Analysis

Patient adherence was calculated as the number of days with one or with two measurement sessions divided by the number of days on which the patient was able to perform the measurements. The agreement between home and hospital measurements obtained on the same or on two consecutive days was assessed by the method of Bland and Altman (13). Individual baselines and coefficients of variation (CV) were calculated once a week for FEV₁ and FEF_25-75 from data obtained during the previous week (or the previous 2 wk if the number of measurements was < 10), and the individual value for a significant change (VSC) was calculated as CV × 1.65 (14). Recalculation of the baseline and cutoff values (baseline minus VSC) for FEV₁ and FEF_25-75 once a week was done to take into account progressive decreases or increases in pulmonary function. Three consecutive sessions showing a decrease of one or both variables from baseline that exceeded the personal VSC generated an alarm, which led the transplant physician to call the patient. Patients who reported new respiratory symptoms or temperature were asked to come to the hospital to answer a standardized symptom questionnaire, perform spirometry, undergo physical examination, and, if appropriate, fiberoptic bronchoscopy with TBB and/or BAL, which were used as gold standards. Only findings on BAL/TBB that deserved treatment were considered clinically significant. If there was an obvious explanation for the functional degradation (e.g., a significant bleeding during a preceding TBB procedure or a recent rib fracture), BAL/TBB was not performed. Similarly, if the patient did not report a new symptom, home pulmonary function was closely followed, and the patient was contacted to perform additional procedures only if the functional decline persisted for more than 1 d. The alarm episodes were classified as true positives (TP), false positives (FP), and false negatives (FN). Data are expressed as mean ± SD.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The median follow-up was 473 d (range, 60-822), during which a total of 13,833 tests were performed (median, 683 per patient; range, 105-1,343). Two patients died during the course of the study, 1 of cytomegalovirus pneumonia and 1 of bronchiolitis obliterans syndrome (BOS) (15); at the end of the study, one patient was in BOS stage 1. Average patient adherence to the monitoring was 55 ± 21% (range, 13-92%) for two measurement sessions a day and 85 ± 10% (range, 65-100%) for one measurement session a day. Figure 1 illustrates that patient adherence tended to decrease with posttransplant time; for the 13 patients who were followed for at least 1 yr, adherence for two measurement sessions a day averaged 72.2% during the first 6 mo and 61% during the last 6 mo (p = 0.018 by paired t test); corresponding figures for one measurement session a day were 93.3% and 89.9%, respectively (p > 0.05).

View larger version (48K): [in this window] [in a new window]   Figure 1.   Average values for patient adherence with the monitoring for two (closed bars) and one (open bars) measurement sessions a day as a function of follow-up time. The figures on top of the bars refer to the number of patients studied at each timepoint. Note the gradual decline in adherence over time.

Figure 2 shows the comparison of home and hospital values of FEV₁ and FEF_25-75 recorded on the same (±1) day in 22 patients on 499 occasions. Home values of FEV₁ tended to be slightly smaller than hospital values with a mean difference of 114 ml (3.8%), whereas the converse was true for FEF_25-75, with a mean difference of +206 ml/s (+5.9%). These small differences might be related to procedural issues (e.g., the aid of a trained technician was provided for hospital but not for home measurements, and home and hospital measurements were not obtained at the same time of the day) and to differences between instruments. Measurements obtained in four normal subjects with the two spirometers placed in series demonstrated a mean difference (MicroloopSensormedics) of +37 ml (1.1%) for FEV₁ (p > 0.05) and +83 ml (2.4%) for FEF_25-75 (p = 0.022).

View larger version (33K): [in this window] [in a new window]   Figure 2.   Comparison of home and hospital values of FEV1 (top panel) and FEF25-75 (bottom panel) obtained on the same (±1) day on 499 occasions in 22 lung transplant recipients. The difference between home and hospital values is on the y axis and the average of home and hospital values is on the x axis; dashed lines indicate the limits of the 95% confidence interval, and dotted lines indicate the mean difference for all data points.

Table 2 shows the weighted mean CV and VSC for several functional variables; a weighted average was used to correct for the effect of an unequal number of measurements between patients. FEV₁ had the lowest CV and VSC, FVC and peak expiratory flow rate had intermediate values, and the highest values were observed for maximum expiratory flow rates.

Over the study period, a total of 103 alarms were generated by 21 patients; of these alarms, 40 were classified as TP and 63 as FP. Table 3 shows that 20% of all TP were detected by a change in FEV₁ alone, 23% by a change in FEF_25-75 alone, and 58% by simultaneous changes in the two variables; on average, the drop from baseline in FEV₁ and FEF_25-75 amounted to 10.9 ± 8.0% and 26.3 ± 14.4%, respectively. The 40 TP occurred in 12 patients and were related to one or more of the following complications: respiratory infections (n = 25), acute rejection (n = 12), and miscellaneous causes (a transient drop in lung function following a BAL/TBB in four patients, and a rib fracture in two patients). Sixty percent of all TP were accompanied by symptoms, whereas 40% were asymptomatic. On 63 occasions, 21 patients generated an alarm for which no specific diagnosis could be found (FP), and on each occasion the drop in lung function recovered without specific treatment (Table 3); 48% of these alarms were unique, that is, on the next session, the variable that had generated the alarm was back to baseline (±VSC). Finally, 24 FN were identified in 12 patients on BAL/TBB, which were performed either because the patient reported symptoms (33%) or as part of the routine surveillance protocol; these 24 FN consisted of 14 episodes of infection and 10 episodes of rejection. Based on these figures, the sensitivity and PPV of combined measurements of FEV₁ and FEF_25-75 for the detection of acute allograft dysfunction were 63% and 39%, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The system used in the present study proved to be technically feasible and was well accepted by all patients but one. The average adherence to the monitoring was 55% for two measurement sessions a day and 84% for one measurement session a day. In a study of 41 HLT, BLT, or SLT recipients who were followed for 1 yr using a paperless electronic spirometer, Finkelstein and colleagues (8) reported that on average 82% of the patients transmitted functional data once a week to the transplant center via a modem. This percentage decreased from above 90% during the first 8 wk of the study to 70 to 80% subsequently. The average number of measurements per patient transmitted each week also decreased from about six to three to four (for an expected number of seven). In our study, patient adherence decreased by ~10% for two measurement sessions a day between Months 1-6 and Months 7- 12 in the 13 patients who were followed for at least 1 yr. For the corresponding periods, the average number of measurements per patient and per week decreased from 12.6 to 11.2 (for an expected number of 14). So the effect of posttransplant time on patient adherence was very similar in the present study and the study of Finkelstein and colleagues (8), although these authors developed a comprehensive program designed to increase patient adherence (16) (e.g., an automatic response tracking/calling system in case of missing data and a regular mailing of newsletters and review graphs of individual spirometric data).

Overall agreement between home and hospital measurements was very satisfactory for both FEV₁ and FEF_25-75. The 114 ml difference in FEV₁ found here in 22 patients who provided 499 pairs of data is lower than the 222 ml difference initially reported in 16 patients by Finkelstein and coworkers (7), but it is similar to the 120 ml difference subsequently reported by the same group of investigators in a larger group of patients (65 patients; 382 pairs) (9). The good agreement found here might be explained by at least two factors. First, the home computer provided the patient with a real-time visual feedback of the flow-volume curve, which presumably worked as an incentive to improve the quality of the expiratory maneuvers. In addition, the computer asked the patient to repeat the maneuver whenever the FEV₁ was below the personal reference value, which increased the chance of obtaining truly maximum achievable values.

In 77 lung transplant recipients who were followed at home during an average period of 441 d using a paperless electronic spirometer, Lindgren and coworkers (9) reported a median standard deviation for FEV₁ of 0.04, which compares well with the value of 0.09 found in the present study. The latter value corresponded to a VSC of 5.2%, whereas the VSC for FEF_25-75 averaged 12.3% (Table 2). These VSC are ~50% smaller than those previously reported by Martinez and coworkers (17) and Estenne and coworkers (18) for in-hospital measurements obtained in HLT and BLT recipients. It thus appears that by performing spirometry on a daily basis, lung transplant recipients may become very reproducible in their functional measurements, not only for FEV₁ but also for FEF_25-75, which in the past has often been considered too variable to be clinically useful.

It is unclear why the Internet-based system for home lung function monitoring used here achieved a sensitivity of only 63%, which is comparable to the value previously reported for hospital measurements of FEV₁ (1). The fact that home spirometry was measured daily and included both FEV₁ and FEF_25-75 and the relatively low VSC for both variables (Table 2) were indeed expected to make the current monitoring more sensitive than hospital spirometry. As shown in Table 3, recording FEF_25-75 did actually increase the number of TP by 23% by detecting events that would not have been detected by FEV₁ alone, but the number of FN was still relatively elevated compared with the number of TP.

Analysis of FN indicated that only one of these events could be directly attributed to updating of the baseline toward a lower value. The FN were equally distributed among patients with and without TP, that is, we could not identify a population of responders (with many TP and few FN) and nonresponders (with few TP and many FN). An attractive hypothesis to account for this observation would be that the functional response to acute infection and rejection is determined, at least in part, by the presence of nonspecific bronchial hyperreactivity (NSBHR): because the degree of NSBHR varies substantially over time (19), a patient may be a responder at a given timepoint after surgery and a nonresponder at another timepoint. This might explain why home spirometry did not do better than hospital spirometry, that is, compared with hospital spirometry, home spirometry would help detect functional changes at an earlier stage in the responders, but it would not be beneficial in the nonresponders. The presence of NSBHR might also account for the transient fluctuations in lung function for which no specific cause was found (5); in the present study, these fluctuations produced a substantial number of FP, and hence decreased the PPV of home spirometry.

These results thus clearly indicate that close monitoring of FEV₁ and FEF_25-75 fails to detect a significant proportion of episodes of allograft infection and rejection; the recommendation made by some centers to stop performing surveillance BAL/TBB should therefore be reevaluated very carefully. The relatively low sensitivity of home lung function monitoring, however, should not be interpreted as suggesting that it is not clinically useful. On several occasions, it detected complications that were otherwise totally asymptomatic; in addition, monitoring lung function proved useful in following patients who developed nonspecific symptoms, and it provided an easy means to monitor the functional response to treatments given at home. These aspects may be particularly important in larger countries where patients live at a long distance from the transplant center.

In conclusion, home monitoring of FEV₁ and FEF_25-75 via the Internet in lung transplant recipients is feasible and well accepted by the patients. The agreement between home and hospital spirometry is very good, and the intrasubject coefficients of variation are low. Yet home spirometry has a sensitivity of only 63% for the detection of acute allograft dysfunction, which is comparable to hospital spirometry. This reflects the fact that a substantial proportion of acute complications do not alter pulmonary function.