Association between dementia and mortality in the elderly patients undergoing hip fracture surgery: a meta-analysis

We searched PubMed, Embase, and Web of Science until April, 2018, without language restrictions. In addition, the references of the included studies were manual searched to identify any additional articles. The following keywords were adopted in the database search: “hip fracture,” “mortality,” and “dementia.” The Boolean operators were used to combine them.

The inclusion criteria are as follows: (1) studies examined the relationship between dementia and mortality among elderly patients undergoing hip fracture surgery; (2) the sample size was more than 200; (3) study designs were cohort study; (4) the mean age is greater than 60 years. The exclusion criteria are as follows: (1) other types of fractures included; (2) studies provided insufficient data; (3) case report, review, commentary, and study just included an abstract; (4) dementia with other psychiatric disorders.

Two reviewers (JZ B and P Z) independently extracted the following information from each study: first author, year of publication, country, study design, sample size, mean age, diagnostic criteria for dementia, and duration of follow-up (in-hospital, perioperative setting, 30-day, 6-month, 1-year and more than 1-year). Any discrepancies were resolved following discussion. All extracted data were entered into a predefined standardized Excel (Microsoft Corporation, USA) file carefully.

We evaluated the quality of studies by Newcastle-Ottawa Scale (NOS), which is a risk of bias assessment tool for observational studies [4]. This scale contains a number of answers per question ranging from two to five. High-quality responses earn a star, totaling up to nine stars. We set scores of 0–3, 4–6, and 7–9 for low, moderate, and high quality of studies, respectively [5].

Statistical analyses were performed by using STATA 14.0 (StataCorp LP). The association between dementia and mortality was described as risk ratios (RR) with 95% confidence intervals (CI) by using random models. P ≤ 0.05 was regarded as statistically significant. Statistical heterogeneity was assessed by the Cochran Q test and quantified by the I² statistic [6], and an I² value greater than 50% indicates significant heterogeneity. When significant heterogeneity existed, heterogeneity analysis would be conducted to make a further explanation. We divided the follow-up time into four timepoints: 30-day, 6-month, 1-year, and more than 1-year follow-up. We classified the in-hospital and perioperative mortality as 30-day mortality.

Two reviewers (JZ B and P Z) independently undertook the searches. Endnote X8 (version 18.0.0.10063) was used to remove duplicate studies. Additionally, we deleted irrelevant articles through the full text. Finally, 18 articles were included. The study selection process was shown in Fig. 1, and the main characteristics of the included trials were summarized in Table 1. Fourteen of 18 were prospective cohort studies [7‐20], and four were retrospective cohort studies [21‐24]. These studies enrolling 295,285 subjects were published between 2000 and 2018, with a sample size ranging from 272 to 134,144. The average scores of the quality assessment were 7.3 (range, 7–9). Details of the quality assessment were available in Table 2.

Seven studies [7, 13, 15, 18, 19, 21, 24] provided available data, and the pooled results suggested that dementia significantly increased the mortality of patients undergoing hip fracture surgery [RR = 1.57, 95% CI (1.29, 1.90), P < 0.001, I² = 23.4%; Fig. 2], and the mortality of patients with dementia suffered from hip fracture surgery [ES = 12%, 95% CI (8%, 15%), I² = 77.5%; Fig. 3].

Only two studies [8, 11] provided available data in 6-month follow-up. The pooled outcomes suggested that dementia significantly increases the mortality of patients after hip fracture surgery [RR = 1.97, 95% CI (1.47, 2.63), P > 0.001, I² = 41.8%; Fig. 2], and the mortality of the dementia patients suffered from hip fracture surgery [ES = 32%, 95% CI (17%, 48%), I² = 91.8%; Fig. 3].

Seven studies [8, 10, 13, 14, 17, 22, 23] provided available data, and the pooled results suggested that dementia significantly increases the mortality of patients undergoing hip fracture surgery in 1-year follow up[RR = 1.77, 95% CI (1.54, 2.04), P < 0.001, I² = 91.7%; Fig. 2], and the mortality of the dementia patients suffered from hip fracture surgery [ES = 39%, 95% CI (35%, 43%), I² = 89.8%; Fig. 3].

Five studies [8, 9, 12, 16, 20] reported more than 1-year mortality. The pooled outcomes suggested that dementia significantly increases the mortality of patients undergoing hip fracture surgery [RR = 1.60, 95% CI (1.30, 1.96), P < 0.001, I² = 87%; Fig. 2], and the mortality of the dementia patients suffered from hip fracture surgery [ES = 45%, 95% CI (32%, 58%), I² = 97.3%; Fig. 3].

Our meta-analysis demonstrated that the mortality of patients with dementia suffered from hip fracture surgery is 12%, 32%, 39%, and 45%, and dementia increased 1.57, 1.97, 1.77, and 1.60-fold mortality in patients undergoing hip fracture surgery in 30-day, 6-month, 1-year, and more than 1-year follow-up. A previous meta-analysis included 75 studies, and involving 64,316 patients [25], the pooled outcomes demonstrated that the overall inpatient or 1 month mortality was 13.3%, 3 to 6 months was 15.8%, 1 year 24.5%, and 2 years 34.5% in generally patients, which mortality was significantly lower than patients with dementia who underwent hip fracture surgery.

To the best of our knowledge, there are three meta-analyses that described risk factors for postoperative mortality after hip fracture surgery [25‐27], which dementia is one of these risk factors. However, these meta-analyses included fewer literatures for the indicator of dementia, and inconsistent follow-up time made conclusions less convincing. Therefore, we performed an update meta-analysis to evaluate the effects of dementia on postoperative mortality of patients undergoing hip fracture surgery and the incidence of death in patients with dementia undergoing hip fracture surgery in 30 days, 6 months, 1 year, and more than 1 year follow-up.

Dementia has not received enough attention compared with deep vein thrombosis and pulmonary embolism in patients with hip fracture. However, our pooled results demonstrated that more than one third of people with dementia will die after hip fracture surgery in a 1-year follow-up and about one in two in more than 1-year follow-up. Patients with dementia usually have less activity and poor self-care ability, which increased postoperative complications, such as surgical site infection, urinary tract infection, and respiratory complications [28]. Therefore, orthopedists should attach great importance to this age-related disease, instead of just focusing on hip fracture problems. Besides, they should fully assess whether the dementia patients are suitable for surgical treatment after suffering from a hip fracture. Can surgical treatment improve the quality of life and prolong life span of patients, especially for patients with severe dementia? For patients with dementia who underwent hip fracture surgery, we should establish a multidisciplinary model of diagnosis and treatment to improve the postoperative outcomes [29]. Besides, nursing cares are necessary to prevent the adverse outcomes of dementia postoperatively [30], such as aspiration, pressure sore, deep vein thrombosis, and pneumonia. Further, rehabilitate exercises are needed to restore function as much as possible after discharge.

Our results show that patients with dementia have very high mortality rates after undergoing hip fracture surgery. Mortality will be higher for patients with severe dementia. It is unclear whether surgical treatment could improve quality of life and prolong life-span of patients with severe dementia. Until recently, there is no relevant meta-analysis or clinical guidelines to elaborate on this issue. We recommend that future studies should focus on which one is better for the treatment of patients with severe dementia who suffered from hip fracture, surgery, or conservative treatment. Besides, clinical decision makers should formulate relevant clinical guidelines to guide clinicians in the treatment of patients with dementia suffered from hip fracture.

The pooled results indicated that significant heterogeneity was found in some indicators. For the prevalence of death of patients with dementia after hip fracture surgery, (1) we conducted a sensitivity analysis through the deletion of Jantzen et al’s study [13], and the heterogeneity was reduced from 77.5 to 0.0% in 30-day mortality; (2) we excluded each study of the outcomes one at a time in 1-year mortality, but the heterogeneity was still very high. Unequal levels of regional medical care and significant sample size difference in the included studies may be the major sources of heterogeneity. For the effect of dementia on postoperative mortality, there was significant heterogeneity in 1-year mortality. The heterogeneity was reduced from 91.7 to 33.8% after the removal of Mitchell et al’s study [22]. Only two studies were included in 6-month mortality, which may be the main source of heterogeneity. In terms of more than 1-year mortality, the inconsistency of follow-up time and obvious sample size differences, which may be major sources of higher heterogeneity.

The major strengths of this study were the following: (1) our study was restricted to the elderly dementia patients undergoing hip fracture surgery, which were higher homogeneous and selective; (2) we designed our meta-analysis into four timepoints: 30-day, 6-month, 1-year, and more than 1-year follow up, and follow-up time is consistent; (3) we calculated the mortality of patients with dementia suffered from hip fracture surgery.

Limitations of this meta-analysis must be considered. First, although the patients we included in the study were elderly patients, it is difficult to make the confounding variables among the studies exactly the same. Second, we did not classify dementia into mild, moderate, and severe dementia because no related data were available in the original studies. Third, there was a great difference in the sample size of the included studies, which may affect the accuracy of our results to some extent. Fourth, significant heterogeneity was found in some outcomes. Last, although we conducted a comprehensive search of literature, it was hard to rule out the existence of publication bias.