The incidence and prevalence of osteoarthritis (OA) increase with aging of the population, particularly in industrialised countries [1]. Osteoarthritis of the lower limbs can seriously affect subject’s independence and thereby influence lifestyle and care in old and very old patients.

The major rheumatological scholarly organisations (EULAR, OARSI, ACR) have published recommendations on the treatment of knee and hip OA [2-5]: all emphasize the required combination of pharmacological and non-pharmacological measures, including aerobic and resistance exercises, aquatic exercises, and weight loss in overweight or obese subjects. However, these recommendations are based on clinical trials conducted in patients with an average age between 60 and 70 years, therefore very different from very old and/or polypathologic patients, and are not scaled according to patient’s age and physical condition.

The American Geriatrics Society (AGS) published in 2001 recommendations for prescribing physical exercise in old patients [6]. However, they are based on relatively ancient trials and not differentiated according to age or polypathology.

EULAR 2013 recommendations [7], which give detailed guidance for the non-pharmacological management of hip and knee OA, do not take into account the age of patients, nor the feasibility or hazards of exercise and weight loss in old and very old patients.

A group of geriatrician-rheumatologists was formed a few years ago within the French Society of Rheumatology (SFR - Société Française de Rhumatologie): the Geriatrics and Rheumatology Association (AGR - Association Gériatrie et Rhumatologie). More recently a wider working group entitled AGRHUM was formed within the French Society of Geriatrics and Gerontology (SFGG - Société Française de Gériatrie et Gérontologie). These groups decided to take an inventory of existing clinical trials on the treatment of OA in very old patients. We present here this systematic review of trials on physical exercise and weight loss in hip and knee OA in very old patients (75 years and older).

We performed a systematic bibliographical search of the literature looking for clinical trials on non-pharmacological treatment of hip and knee OA in patients aged 75 years and older. The search was focused on physical exercise, aquatic exercise and balneotherapy, and weight loss. Education and information, physiotherapy (hot/cold, laser, ultrasound, percutaneous electrical stimulation), massage, Spa-therapy and acupuncture did not fall within the scope of this research.

The search strategy was as follows:

1. Search in MEDLINE, EMBASE, and PASCAL database for manuscripts published in English, using the following keywords: hip/knee OA; exercise; aquatic exercise; balneotherapy; weight loss; weight reduction. The activated filters were: clinical trials, abstract available, published in the last ten years, age eighty and over. For this last age filter, we initially selected a "sixty five and over" filter: this filter, active as soon as age ≥ 65 years was quoted in the text, resulted in the case of osteoarthritis, to make no selection for age. This led to use the "eighty and over " filter, which is activated as soon as age ≥ 80 years was quoted in the text.
2. Study of the Cochrane Reviews: “aquatic exercise for the treatment of knee and hip osteoarthritis”, “exercise for osteoarthritis of the hip”, “exercise for osteoarthritis of the knee”, “home versus center based physical activity programs in older adults”, ”intensity of exercise for the treatment of osteoarthritis”, ” balneotherapy for osteoarthritis”. In each of these reviews a comprehensive manual search of selected trials and excluded trials was performed.
3. Manual search of references in international guidelines, meta-analyses and systematic reviews.
4. Manual search of references in all relevant articles obtained during the previous steps.

We identified a total of 83 trials (Table 1).

Among them, only two concerned subjects aged ≥ 75 years on average; we therefore decided to lower the mean age for selecting articles to 70 years. This led to a selection of 15 papers and 13 trials [8-22], two being duplicates [8, 9, 14, 15]. The 2 trials on weight loss [23, 24] included patients with a mean age slightly under 70 years: we however, kept them, given the scarcity of trials on weight loss in elderly patients. The methodological quality of these 15 trials, was assessed by the Jadad score [25], with an average score of 3/5.

Heterogeneity of trials did not allow carrying out a meta-analysis.

To our knowledge, this is the first systematic review of the literature on the tolerance and effectiveness of physical exercise and weight loss for hip and knee OA in very old patients, which are first line therapeutic options recommended by all published therapeutic guidelines in OA [2-7].

- Physical exercises were found to be effective on pain and function. These results are highly significant given that such trials are subject to strong methodological constraints: single-blind is sometimes difficult to enforce [8], and low sample sizes frequently limit statistical power, which in additioncould also be reduced by the possible effect of sham treatment in the control group: brochures and educational sessions may encourage controls to increase their activity by themselves, and trips to the centre can provide general stimulation [13]. In the literature, effectiveness of physical exercise in the elderly is considered equivalent to that observed in younger adults [6]. The negative results of the trial by Williamson [16] can be explained by two limitations: subjects included with very advanced osteoarthritis and the low number of sessions, 6 performed at a weekly interval. This trial is the one with the lowest number of sessions, whereas a dose effect has been suspected [12]: in adults, a Cochrane Review on exercises for knee osteoarthritis [26] concluded that programmes including at least twelve sessions are more effective than those with less.

- Most trials on aquatic exercises found a short-term effectiveness. The population studied in these trials is deemed independent, but some subjects are nevertheless on waiting lists for arthroplasty [18, 20], and therefore suffering from advanced osteoarthritis; however aquatic exercises are effective and well tolerated in these patients. The negative results of the Hale’s trial [22] could be explained by a lack of power, and a possible positive effect of centre-based computer-training sessions in the control group.

- The small number of trials found on weight loss precludes any formal conclusion. Weight loss alone seems ineffective on pain and function [23]. Diet + exercise combination seemed effective [23, 24], even in case of moderate weight loss (5-8% of the initial weight).

- The beneficial effect of muscular training on osteoarthritic joint may improves joint stability, and potentially slows down joint degradation [13, 27] ; it also prevents stiffness and articular flessum worsening [28]. Among patients suffering from knee osteoarthritis, isometric quadriceps exercises have been shown to increase hyaluronic acid molecular weight and synovial fluid viscosity [29].

- Whether aerobic or resistance training (anaerobic) should be recommended, is still debated. A meta-analysis in adults [30, 31] found resistance exercises effective on short-term pain, and aerobic exercises effective on long-term function. Resistance exercises have not been enough studied in elderly [32]. Both kind of exercises may be complementary and should be included in rehabilitation programmes for OA patients [14, 30, 31].

- The mechanism of action of aquatic exercises does not seem different from that of land-based exercises. The analgesic effect of hot water may explain short-term effectiveness on pain, and good tolerance of intensive aquatic exercises [20]; immersion decreases the muscular work required for joint mobilisation and allows to take care of patients who do not tolerate land-based exercises [19]; nevertheless it probably reduces the gain of muscle strength obtained [18]; partial immersion also reduces the risk of falls, which is interesting in patients with an unstable equilibrium [22].

- Intensive weight-loss diet reduces pain and improves function in adults with knee osteoarthritis through overall weight loss [33]. It modifies also body composition by reducing body fat mass, but also lean body mass in equivalent proportions to those observed in adults; this decrease in lean body mass does not prevent improvement in function at six months [24] but long-term data are missing.

- Overall, safety was good. Aerobic and resistance exercises used in trials were adapted for old patients and often gradually implemented, as recommended [6]. Weight loss-induced sarcopenia and osteoporosis must be carefully considered before prescribing an intensive and/or prolonged diet in old patients [34], even if intentional weight loss is not associated with increased mortality in old subjects [35].

Land-based exercises should be performed between 1 and 3 times a week; session duration depends on the type of exercise, from 25 minutes on a bike to 90 minutes if associated rehabilitation. The duration of exercises should not exceed 60 minutes per session, and is generally gradual, to improve tolerance, compliance and effectiveness [12].

Centre-based rehabilitation stimulates patients by encouraging travelling and meetings: this may explain improvement observed in some control groups. Nevertheless it requires from patients several trips a week, and is difficult to convert into a home-based maintenance programme. Home-based rehabilitation [12] is more difficult to initiate and to assess for compliance and effectiveness, but its long-term maintenance seems better. A Cochrane Review [36] favoured home-based over centre-based rehabilitation in the elderly, but none of the trials were performed in OA.

Whether centre-based or home-based, the persistence of physical activity remains unsolved in OA patients, regardless of age: maintenance can be improved by implementing “booster sessions” [37], focusing on exercise requiring costless equipment [14], and by a customised strategy of positive reinforcement [30].

- The good safety of aquatic exercises in subjects with advanced OA [18] is of particular interest. However aquatic rehabilitation requires a higher infrastructure and financial investment than land-based rehabilitation. Aquatic exercises are therefore inadequate for patients with dementia or polypathology. In the absence of demonstrated superiority of aquatic rehabilitation, land-based rehabilitation should be the first recommended option [20].

- The only study performed on institutionalised elderly patients does not report long-term results [17]. Institutionalised subjects who benefit from sessions in their institution should have a more sustainable result if a suitable maintenance programme is settled. A recent Cochrane Review [38] on rehabilitation for institutionalised older people analysed 67 trials involving 6,300 patients (average age 83 ± 8 years), without focusing on OA. It concluded that rehabilitation modestly but significantly improves independence of residents, and their muscular strength, range of motion and balance, and was well tolerated. This regards only half of the residents, a major cognitive decline being a major contra-indication, but can successfully be practised by residents moving with technical assistance.

- We found 13 trials with an average age ≥ 70 years, which means the presence of subjects aged between 50 and 70. Collection of each patient’ data would have allowed us to select and analyse separately subjects over 75 years: we were not able to access database.

- Few trials were conducted in hip OA. Furthermore, trials including hip and knee OA did not generally stratify analyses according to the joint affected.

- To complete our review, we also searched for ongoing trials on non-pharmacological treatment in hip and knee OA on “ClinicalTrials.gov” website. For land-based exercises, we found fourteen trials, including one described in this review [17]. Of the remaining thirteen trials, only one focused on hip OA; three had a lower age limit above 60 (62, 65 and 70 years). For aquatic exercises, we only found one trial, included in this review [21]. For weight loss, the site reported only the Messier trial [23].

This orientation of research toward land-based exercises seems realistic for very old or polypathologic subjects. However 8 of these 13 trials a priori excluded patients suffering from dementia and/or polypathology, and none was performed in an institution.

This systematic review confirms that international recommendations for the non-pharmacological treatment of knee OA also apply to subjects aged 70 - 80 years. Evidence for hip OA is insufficient. The complementary nature of land-based and aquatic exercises, and the need for reinforcement strategies to sustain physical activity must be underscored.

We also confirm the lack of published trials and research projects on the risk/benefit ratio of such interventions in very old and/or polypathologic subjects. Specific trials focused on land-based exercise, particularly in institutionalised subjects are needed. Performing such trials is a major public health challenge in these populations exposed to drug-related iatrogenesis.

Based on this review, our recommendations are to propose to all old patients at least regular land-based exercises sessions, if not painful on moves, in order to improve mobility and muscle strength (2 to 3 times a week, sessions of 20 to 60 minutes according to the type of exercise). Additionally, managing a weight loss in overweight patients might be useful, after ensuring they are not affected by sarcopenia.