In 2005, all 201 patients who according to the physical therapy records of Kymenlaakso Central Hospital had undergone unilateral knee replacement 4 to 18 months prior to the study were informed about the study. Eighty-six patients responded and were contacted by the research personnel and interviewed over the telephone. Patients with bilateral knee arthroplasty, revision arthroplasty, severe cardiovascular diseases, dementia, rheumatoid arthritis, or any major surgery in either of the knees were excluded from the study. Thus, 50 eligible volunteers (age range, 55–75y), 30 women and 20 men, were randomly assigned after the baseline measurements into an aquatic resistance training group (16 women, 10 men) and a control group (14 women, 10 men). The random allocation was concealed in sealed envelopes in blocks of sex, age, and type of knee replacement. The intervention profile is displayed in figure 1.

Fig 1

Intervention profile.

 

The reason for the knee replacement surgery for all the participants was knee joint osteoarthritis. Details of the knee replacement operation were collected from the hospital medical records. In all cases, the knee replacement surgery had been performed with cement fixation (48 with tricompartmental total knee arthroplasty, 2 with unicompartmental hemiarthroplasty). The participants with hemiarthroplasty did not differ from those with total knee replacement in any of the variables.

Before the laboratory examinations, the participants were informed about the study, and they gave their written informed consent. The study was conducted according to the Declaration of Helsinki and approved by the ethical committee of Kymenlaakso Central Hospital.

Quantitative CT measurements and analyses were conducted blind to the study group. The other measurements were conducted unblind.

The general health, clinical history, medication, and diseases of the participants were assessed by a physician before the laboratory examinations to evaluate potential contraindications for safe participation in the measurements and training. Body height and weight were measured in the laboratory using standard procedures. The presence of self-reported chronic conditions was recorded by a questionnaire. Habitual physical activity was recorded during the intervention using a training diary.

Maximal³¹ and habitual³² walking speed over 10m were measured in the hospital corridor, and the time taken was recorded using photocells.^a First, the participants were instructed to walk at their habitual walking speed. Second, they were instructed to walk as fast as possible without compromising their safety. All the participants wore thin aquatic shoes and were allowed 3m for acceleration. Each participant performed 2 trials at maximal and 2 at habitual walking speed separated by a 1-minute rest, and the faster performances were accepted as the results. In our laboratory, the ICC for persons with knee replacement has been .86 for maximal and .44 for habitual walking speed.¹⁹

The WOMAC questionnaire,³⁴ a self-rated measure of pain and stiffness and the physical functional difficulty of the participants, is widely used after joint replacement surgery research.^{29, 35} The version based on the visual analog scale (range, 0–100mm, with 100 indicating the worst possible situation) was used. In the physical functional difficulty score, 80% of the participants did not answer the subscale “getting in and out of the bath” because they did not have a bath. Therefore, this subscale was not included in the analysis.

Maximal muscle power of the knee extensors and flexors was measured with an isokinetic dynamometer^b with a sampling frequency of 100Hz and measurement error of 1% throughout the entire range of motion.²⁷ The dynamometer was calibrated before each measurement session according to the procedure recommended by the manufacturer. Before the measurement session, the participants were carefully familiarized with the testing procedure. For each leg, the axis of rotation of the dynamometer was aligned with the condylus femoris lateralis. The lever arm of the dynamometer was attached around the ankle 2.5cm above the midpoint of the malleolus lateralis. Hip and thigh were stabilized with straps. The measurement was performed on as large a range of motion of the knee as possible. The nonoperated leg was measured first. After 2 to 3 submaximal flexion-extension movements, 5 maximal continuous flexion-extension trials were performed at an angular velocity of 180°/s. The participants were verbally encouraged to make a maximal effort throughout the whole range of motion. Peak knee extensor and flexor power (W) values were analyzed from the best extension and flexion efforts. The ICC of the isokinetic parameters for persons with knee replacement was between .90 and .96 for the operated knee.¹⁹

Exercise sessions were conducted twice a week in small classes containing 4 to 5 persons. All the classes were supervised by an experienced physiotherapist. Table 1 summarizes the training program, including the weekly sets, duration of work, rest, and the training load produced by the resistance boots as well as the mean RPE value for each week.

Each session started with an 8-minute warm-up including walking (forward, backward, and sideways), aqua jogging, and lower-leg muscle stretching. This was followed by 30 to 40 minutes of resistance training and a 5-minute cooling down period. The exercises were selected on the basis of our previous study.²⁷ Each training session consisted of 5 exercises for both legs: (1) knee extension-flexion movement in a sitting position, (2) hip abduction-adduction with extended knee in a standing position, (3) hip extension-flexion with extended knee in a standing position, (4) knee extension-flexion in a standing position, and (5) step-squat backward from the aqua aerobic step board. The subjects were verbally encouraged to perform each repetition with maximal effort in order to achieve the highest possible movement velocity and resistance. In each exercise, the operated leg was trained first and then the nonoperated leg. The operated leg was trained to 30% more sets compared with the nonoperated one. The participants were asked to describe their perceived exertion after each exercise with the RPE scale (range, 6–20).³⁷ The participants were asked to report whether they had any pain or discomfort during the training sessions.

The progression of the exercise program was ensured by using resistance boots of different sizes and by varying the amount and duration of sets. The first 2 weeks of training were conducted without resistance boots in order to adapt to the exercises. The actual training was conducted using small Aqua Runner Zero Impact Footwear^e (2 weeks) and with medium (4 weeks) and large resistance boots^f (4 weeks). In weeks 7 and 12, one training session was conducted without resistance boots in order to avoid overtraining. The frontal area of the medium size resistance boots was 0.045m² and that of the large resistance boots, 0.075m². The small footwear was attached around the foot and the medium and large boots around the lower leg and foot during the exercises. In our previous study,²⁷ the drag during the exercises in healthy women was double with the medium boots and triple with the large boots compared with the barefoot condition.

Intensity of training was estimated for 3 women and 3 men in the training group (mean age ± SD, 62.2±4.3y; mean height ± SD, 169.8±8.2cm; mean weight ± SD, 86.3±9.2kg) by the RPE scale and with heart rate monitoring. The average heart rates were recorded with the Polar RS400 heart rate monitor^g during the 5 exercises, excluding the warm-up and cool-down. Age-related maximal heart rates were calculated according to the following equation: 220–age (y). Among the 6 persons tested for training intensity, the mean RPE value ± SD during the exercises was 17±1 (range, 14–18). The mean heart rate ± SD was 116±18 beats/s (range, 93–148), which was 73% of the age-related maximal heart rate. In the hydroboot conditions, the movement velocities were slower and the number of repetitions a set lower but the resistance higher compared with the barefoot condition.^{27, 38}

The control group did not receive any intervention. Participants were encouraged to continue their lives as usual and maintain their habitual level of physical activity during the trial.

Table 2 shows the baseline physical characteristics of the training and control groups. No between-group differences were observed at baseline.

Table 2Baseline Physical Characteristics of the Participants in the Aquatic Training and Control Group

Characteristics	Aquatic
	Training Group	Control Group
	n=26	n=24
Age (y)	66.2±6.3	65.7±6.0
Weight (kg)	83.2±15.2	83.9±15.0
Height (cm)	167.3±9.3	169.7±8.2
Time since operation (mo)	9.9±4.7	9.2±4.2
Comorbidities,⁎ n (%)
Cardiovascular	15 (58)	9 (38)
Endocrine	2 (8)	2 (8)
Musculoskeletal	13 (50)	8 (33)
Respiratory	2 (8)	1 (4)
Diagnosed knee osteoarthritis of the nonoperated knee, n (%)	4 (15)	4 (17)

NOTE. Values are mean ± SD unless indicated otherwise.

Self-reported number of comorbidities.

The dropout rate was 6%. In the training group, 1 participant (knee replacement operation on the other knee), and in the control group, 3 participants (1 for personal reasons and 2 for a knee replacement operation on the other knee) were lost to follow-up (see fig 1). Training compliance in the aquatic training sessions was excellent, averaging 98% (590 sessions attended/600 offered). In the training group, the participants did not report any pain during the training program. However, 1 participant visited the study physician because of elevated blood pressure without taking further actions or a pause in the training regimen. The mean RPE value for training was 16 (range, 14–17).

At baseline, there were no differences between the aquatic training and the control group in maximal walking speed (P=.214), habitual walking speed (P=.684), or stair ascending time (P=.884). In the baseline measurements, 3 participants in the training and 1 in the control group were unable to perform alternate stepping and thus used bipedal ascent.

Compared with controls, the training group showed a mean increase in their habitual walking speed of 9% (95% CI, 3%–15%) at the end of the intervention. Maximal walking speed was not affected by training (1%; 95% CI, −6% to 8%). Stair ascending time decreased significantly among the trainees compared with controls (−15%; 95% CI, −24% to −6%). After the intervention, 1 participant in each group used bipedal ascent.

At baseline, there were no differences between the aquatic training and control group in the WOMAC physical functional difficulty (P=.109), pain (P=.866), or stiffness (P=.254) scores. The scores for physical functional difficulty (analysis of covariance, P=.197), pain (P=.352), and stiffness (P=.097) in the operated knee were not affected by training.