Changes in Cognitive Function From Presurgery to 4 Months Postsurgery in Individuals Undergoing Dysvascular Amputation

Abstract

Objective:

To describe cognition among individuals with new amputations at 3 time points: presurgical, 6 weeks postamputation, and 4 months postamputation.

Design:

Prospective cohort.

Setting:

Medical centers.

Participants:

Referred sample Veterans who were primarily men (N=80) experiencing their first lower extremity amputation as a result of complications of diabetes mellitus or peripheral arterial disease. Patients were screened for the absence of gross cognitive impairment using the Short Portable Mental Status Questionnaire (SPMSQ). Of those 87 individuals who were eligible, 64% enrolled; 29 were enrolled presurgically and have cognitive data for all 3 time points, and 58 were enrolled postamputation. Eighty of the 87 individuals enrolled by 6 weeks remained enrolled at 4 months.

Interventions:

Not applicable.

Main Outcome Measures:

Demographic and general health information, general mental status (SPMSQ), and 4 brief, well-established neuropsychological measures.

Results:

Most mean neuropsychological test scores fell in the low average or average range. For most participants, overall cognitive status improved from pre- to postsurgery and then remained stable between 6 weeks and 4 months. There were significant improvements between pre- and postsurgical test scores in verbal learning and memory, and these remained unchanged between 6 weeks and 4 months. Better 4 month cognitive performance was associated with higher perceived general health.

Conclusions:

Overall cognitive performance is poorest presurgically. Though there is improvement between pre- and postamputation, cognition appears generally stable between 6 weeks and 4 months.

Diabetes mellitus (DM) and peripheral arterial disease (PAD) account for 82% of lower limb amputations in the United States.¹ These chronic diseases are frequently comorbid, with at least 20% of individuals with PAD also having DM² and roughly 25% of individuals with DM also having PAD.^3,4 Among those with lower limb amputation, upward of 48% are diagnosed with DM, and 94% are diagnosed with PAD.⁵

Both PAD and DM can impact cognitive function,^6,7 causing decreases in processing speed, verbal memory, and cognitive flexibility. This has implications for rehabilitation because stronge attention, working memory and visual-constructional skills are associated with successful prosthetic fitting,^8-10 restoring mobility,^8,11-13 and greater independence with activities of daily living.^14,15 Additionally, prevalence of dementia among those with amputations due to dysvascular disease is higher than among amputees without PAD^16,17 (estimates range from 5% to 49%) and is associated with poorer overall rehabilitation outcomes.^18,19

In addition to the underlying disease processes, new amputees may experience postoperative cognitive dysfunction (POCD) or a decline in cognitive performance from pre- to postsurgery related to anesthesia and the surgical intervention. POCD is common 1 week after surgery²⁰ and can persist for as long as 3 months. Late POCD, present beyond 3 months, is more likely to be present among older individuals or those with preexisting cognitive impairment^20-23 and continues to be present in about 1% of older adults at 1 to 2 year follow-up.²⁴ Other superimposed acute factors may create or exacerbate cognitive deficits around the time of amputation surgery. Anxiety and depressive symptoms, which are prevalent among new amputees²⁵ have been associated with decreased cognitive performance,^26,27 as has pain,^28,29 which is reported in 85% of new amputees.³⁰ The presence of even minor infection has also been shown to impact cognitive performance.³¹ A variety of common medications, such as narcotic pain medications, can interfere with cognitive performance as well.³²

Of 3 pertinent studies, all suggest that dysvascular amputees have impairment in several cognitive domains compared with casematched controls or population norms.^10,15,33 Of these, only 1 study is prospective.¹⁵ None of these studies evaluate or describe presurgical cognition.³⁴ This is notable because cognitive status may factor into presurgical rehabilitation decisions (ie, level of amputation).

The overarching objective of this study is to describe cognition from presurgery to 4 months postamputation in individuals undergoing first lower limb amputation. First, we assess whether there are changes in an aggregate measure of cognition over time. Second, we assess variation in specific cognitive domains. Third, we assess if there are subgroups of participants for whom overall cognitive status is more vulnerable to change around amputation.

Methods

Study design

This study is part of a larger multisite prospective cohort study of individuals undergoing major lower extremity amputation surgery because of PAD or DM at 2 Veterans Affairs medical centers, a university hospital, and a level I trauma center. For this study, participants were assessed via in-person interview presurgically and via in-person or telephone interview at 6 weeks and 4 months postsurgically. Participants who were not available presurgically were enrolled at 6 weeks. This study was conducted in accordance with the procedures approved by the human subjects review boards at each study site.

Participants

Subjects were considered eligible if they were age ≥18 years; they were awaiting (or underwent in the last 6wk) a first major lower extremity amputation, defined as transmetatarsal level or higher; and the primary cause of amputation was DM or PAD. Subjects were excluded if they had cognitive or language impairment that would preclude consent or participation, defined by >4 errors on the Short Portable Mental Status Questionnaire (SPMSQ),³⁵ or if they were nonambulatory prior to admission to the hospital for reasons unrelated to impairment of the extremity awaiting amputation, or if the planned amputation was bilateral. Of 239 individuals screened between 2005 and 2008, 136 (57%) met study criteria, and 87 (64%) enrolled (29 presurgically and 58 at 6wk). The primary reason for exclusion was prior amputation (n=34, 38%); these individuals may also have had dementia or failure of the SPMSQ, but this was not assessed once a candidate was excluded for other reasons. An additional 23 (22%) participants were excluded solely because of dementia or failure of the SPMSQ. Five (5.8%) participants died, and 2 (2.3%) were lost to follow-up; 80 (92%) participants remained enrolled at 4 month follow-up (table 1).

Table 1.

Sociodemographic and health data among participants

Characteristic	Enrolled Presurgically (n = 29)	Enrolled Postsurgically (n = 58)
Veteran	27 (96.4)	44 (75.9)
Age (y)	62.77±10.34	61.89±7.91
Female	1 (3.4)	6 (10.3)
Race/ethnicity
White	25 (86.2)	48 (82.8)
Nonwhite	4 (13.8)	10 (17.2)
Average years of education	13.39±1.93	13.55±1.98
Amputation level (6wk)
Below knee (transmetatarsal and transtibial)	25 (86.2)	54 (93.1)
Above knee (transfemoral)	4 (13.8)	4 (6.9)
Meets criteria for MDE (6wk)	5 (19.2)	12 (20.7)
Meets criteria for MDE (4mo)	9 (36.0)	7 (12.7)*
Perceived general health (preamputation)	2.16±1.09	3.34±1.07
Perceived general health (6wk)	3.35±0.75	3.33±1.02
Perceived general health (4mo)	3.24±0.97	3.55±0.88
Body pain present (preamputation)	8 (66.7)	16 (84.2)
Body pain present (6wk)	17 (63.0)	39 (68.4)
Body pain present (4mo)	13 (48.1)	39 (73.6)

NOTE. Values are mean ± SD or n (%). MDE is not available at baseline for those enrolled at 6 weeks postsurgically because we did not ask this question retrospectively. Perceived general health is measured by the Medical Outcomes Study 36-Item Short-Form Health Survey.

^*Those enrolled presurgically reported significantly higher rates of MDE at 4 months postamputation than those who were enrolled postsurgically (χ² = 5.82, P<.05). Age ranged from 47 to 83 years (for those enrolled presurgically) and 48 to 84 years (for those enrolled postsurgically). Perceived general health scores ranged from a possible range of 1 to 5.

Measures

Cognitive tests were administered within the week prior to amputation surgery for 29 subjects. The same battery was administered to 87 subjects at 6 weeks and 80 subjects at 4 months (table 2). There were no significant differences in neuropsychological test scores between those who were enrolled presurgically and those who were enrolled postamputation at either time point, and there were not significant health and demographic difference between the 2 groups.

Table 2.

Mean standardized z scores, SDs, and proportion impaired (<2nd percentile) for individual neuropsychological tests

Neuropsychological Test	Presurgical Baseline			6wk Postsurgery			4mo Postsurgery
	n	Mean ± SD	Impaired, n (%)	n	Mean ± SD	Impaired, n (%)	n	Mean ± SD	Impaired, n (%)
SPMSQ	81	9.21±1.08	0 (0.0)	84	9.39±0.92	0 (0.0)	80	9.38±0.97	1 (1.3)
Digit span total	29	−0.25±0.79	0 (0.0)	85	−0.28±0.84	1 (1.2)	79	−0.11±0.91	0 (0.0)
List learning	28	−1.42±1.27	7 (25.0)	79	−0.94±1.38	14 (17.7)	78	−0.87±1.34	15 (19.2)
List recall	27	−1.00±1.43	8 (29.6)	79	−0.36±1.19	7 (8.9)	78	−0.48±1.22	10 (12.8)
Semantic fluency	29	−0.97±0.79	2 (6.9)	85	−0.52±1.00	2 (2.4)	80	−0.97±0.87	6 (7.5)

NOTE. Scores are z scores except for SPMSQ. The z scores <−2.00 (<2nd percentile) are considered falling into the impaired range, except for the SPMSQ where impaired is classified as scores <6 (>4 errors).

Two rehabilitation psychologists trained research staff to administer the tests and supervised initially to ensure competent administration. Tests were scored by a trained psychometrist and then rescored by a postdoctoral neuropsychology fellow. When scores were discrepant, a project investigator (R.M.W.) reviewed the scores to ascertain accuracy. Standardized z scores were computed using available norms that adjust for sex, age, and in the case of digit span, education also (see table 2). In all cases, a z score of 0 indicates function at the 50th percentile, whereas positive/greater z scores indicate better cognitive performance, and negative/lower z scores indicate poorer function. Scores can be interpreted as follows: average range (−.67≥z≥.67, 25th–75th percentile), low average range (−.66≥z≥−1.33, 9th–24th percentile), borderline (−1.32≥z≥−2.01, 2nd–8th percentile), and impaired (−2≥z, <2nd percentile).³⁶

Neuropsychological tests

Participants completed the 10-item SPMSQ at each time point. Several subtests of the Repeatable Battery for the Assessment of Neuropsychological Status³⁷ were administered to assess executive function (semantic fluency), auditory-verbal learning (list learning), and verbal memory (list recall). The Repeatable Battery for the Assessment of Neuropsychological Status offers alternative forms to minimize practice effects. Form A was used at the presurgical and 4-month assessments, whereas form B was used for the 6-week administration. Participants completed the digit span subtest from the Wechsler Adult Intelligence Scale, 3rd edition³⁸ to assess attention and working memory.

Possible correlates of cognition

Participants were asked about age, sex, years of education, ethnicity, and amputation level (see table 1) at their first assessment. Perceived general health and presence of pain were assessed at all times, and depression was assessed at 6 weeks and 4 months. Though those enrolled presurgically reported significantly higher rates of a major depressive episode (MDE) at 4 months relative to those enrolled postamputation, there were no other significant differences between the 2 groups. Hence, we did not differentiate between subjects based on enrollment time.

Major depressive episode

Because depressed mood can impact cognitive performance,^39,40 presence of an MDE was assessed with the depression module of the Patient Health Questionnaire,⁴¹ a 9-item screening measure. Participants who endorsed ≥5 symptoms (1 of which was depressed mood or anhedonia) on more days than not in the past 2 weeks were coded as having an MDE,¹ and those who did not were coded 0 (no MDE).

Perceived general health

Perceived general health was assessed via a single item: “In general, would you say your health is…?” with response options ranging from 1 (very poor) to 5 (very good). Similar single item measures of health or disability have high levels of validity in a variety of populations and offer the benefit of brevity in administration.^42-44

Presence of pain

At each time point, patients were asked the following: “Do you currently experience any persistent and bothersome pain?” Though there were follow-up items, for the purposes of this article, we simply report the presence or absence of pain.

Data analysis

To describe cognition, we present mean z scores, SDs, and frequency and proportion of scores in the impaired range³⁶ for each neuropsychological test at each time point (see table 2).

To determine if there were changes in overall cognition overtime, we transformed individual z scores such that outliers were rounded up or down so that they fell within a z=±3 (<1st or >99th percentile). Then, an aggregate mean standardized z score was computed for each individual with complete neuropsychological test data at that time point. We conducted paired t tests to compare aggregate mean z scores between presurgical baseline and 6 weeks postamputation, 6 weeks and 4 months postamputation, and presurgical baseline and 4 months postamputation.

To assess changes in specific cognitive domains over time, each of the individual neuropsychological tests was analyzed in a general linear model analysis of variance with time of measurement as a within-subjects factor (table 3). In the general linear model, missing values are deleted on a casewise basis for a within-subjects factor; therefore, our sample sizes for these analyses ranged from 20 to 22. For tests of the main effect of time, we report the Wilks lambda F test. Tests of the within-subjects contrasts are shown for linear and quadratic effects. Because only 20 to 22 subjects had data at all 3 time points, but 70 subjects had data at both 6 weeks and 4 months, we also performed paired sample t tests comparing each neuropsychological test mean score at 6 weeks and 4 months to better ascertain if there were variations among specific domains postsurgery.

Table 3.

General linear model comparing scores for participants with data at all 3 time points

		Presurgical Baseline, Mean ± SD				Within Subjects Contrasts
Neuropsychological Test Score	n		6wk, Mean ± SD	4mo, Mean ± SD	Main Effect of Time, F (df)	Linear, F (df)	Quadratic, F (df)
Digit span total	22	−0.18±0.86	−0.26±0.67	−0.18±0.79	0.16 (2,20)	0.00 (1,21)	0.33 (1,21)
List learning	21	−1.10±1.23*	−0.33±1.09*	−0.52±0.88	5.39 (2,19)†	6.36 (1,20)‡	6.02 (1,20)‡
List recall	20	−0.77±1.51*	−0.27±1.20	−0.17±1.18*	3.67 (2,18)‡	7.62 (1,19)†	0.67 (1,19)
Semantic fluency	22	−0.87±0.84	−0.35±0.96	−0.76±0.57	2.70 (2,20)	0.24 (1,21)	5.67 (1,19)

NOTE. Main effect of time assessed using Wilks lambda.

^*Significant differences in post hoc pairwise comparisons.

^†P≤.01.

^‡P≤.05.

Finally, relations between aggregate mean z scores at each time point and categorical health and demographic variables (ethnicity, presence of pain, sex, amputation level, and MDE assessed at the same time point as each neuropsychological test) were assessed using 1-way analyses of variance. Univariate associations between age, years of education, perceived general health, and aggregate mean z scores were computed using Pearson correlations.

Results

Neuropsychological test scores

In a sample that consisted of predominantly older (mean, 62.2y), Veterans (82.6%) who were men (92%), both the aggregate mean and individual test mean scores fell into the low average (9th–24th percentile, z=−.66 to 1.33) or average (25th–75th percentile, z=−.67 to .67) range at all time points. There were some participants who scored in the impaired range (z<2 or ≤2nd percentile) on each of the tests (see table 2).

Changes in aggregate mean z scores

Twenty participants had complete neuropsychological test data at both presurgery and 6 weeks, 70 participants had data at both 6 weeks and 4 months postamputation, and 23 participants had data at both presurgery and 4 months postamputation (table 4). Overall performance improved between presurgery and 6 weeks (t₁₉=−2.42, P=.03) and between presurgery and 4 months (t₂₂=−1.97, P=.06), but there was no change between 6 weeks and 4 months (t₆₉=1.1, P is not significant).

Table 4.

Comparisons of aggregate mean z scores across time points

Time of assessment	Baseline vs 6 wk (n = 20)	6wk vs 4mo (n = 70)	Baseline vs 4mo (n= 23)
Presurgical baseline	−.71±.89	NA	−.79±.87
6wk postamputation	−.28±.71	−.41±.73	NA
4mo postamputation	NA	−.50±.79	−.52±.68
	t19=−2.42 (P=.03)	t69=1.1 (not significant)	t22=−1.97 (P=.06)

NOTE. Values are mean ± SD or as otherwise indicated. Mean z scores were calculated only for participants with complete neuropsychological test score data at each pair of time points; therefore, sample sizes for calculations vary.

Abbreviation: NA, not applicable.

Changes in specific cognitive domains over time

Participants demonstrated improvement from pre- to postsurgery on 2 of the 4 tests administered (see table 3). The main effect of time on list recall and list learning was significant (F_2,18 = 3.67, P≤.05 and F_2,19 = 5.39, P<.01, respectively). Review of the post hoc contrast scores on list recall indicated improvement in a linear fashion between presurgery (mean ± SD, −.77±1.51) and 4 month follow-up (mean ± SD, −0.17±1.18; F_1,19 = 7.62; P≤.01). Similarly, list learning scores improved significantly between both presurgery (mean ± SD, −1.10±1.23) and 6 week follow-up (mean ± SD, −.33±1.09) and presurgery and 4 month follow-up (mean ± SD, −.52±0.88; F_1,20 = 6.36; P≤.05). In the larger sample, list recall (t₇₀=−1.04, P is not significant) and list learning (t₇₀=.98, P is not significant) scores did not significantly change between 6 weeks and 4 months.

The main effect of time was not significant on digit span or semantic fluency scores; hence, we did not examine post hoc contrasts. Similarly, paired sample t tests among the larger sample comparing these scores between 6 weeks and 4 months revealed no significant differences.

Associations between overall cognitive performance and health/demographic factors

Higher levels of perceived general health at 4 months were correlated with higher aggregate mean z scores at 4 months (r=.38, P≤.01). There were no other significant associations between health and demographic variables and aggregate mean z scores.

Discussion

This sample likely represents the healthiest of persons with amputations related to DM or PAD. Participants were excluded if they had prior amputations, they were nonambulatory prior to admission to the hospital for reasons unrelated to impairment of the extremity awaiting amputation, the planned amputation was bilateral, or if they were unable to pass a coarse mental status screen. Even with these rigorous exclusion criteria, mean performance on neuropsychological tests generally fell in the low average or average range. However, on 3 of the 4 tests administered, some portion of the sample scored in the impaired range³⁶ at each time point.

Overall cognitive performance improved from pre- to postsurgical assessments and remained relatively stable between 6 weeks and 4 months postsurgery. No participants with complete data worsened from pre- to postsurgery. Pre- to postamputation improvements were seen in new learning (list learning) and memory recall (list recall); scores often went from low average prior to surgery to average after surgery. Although the reason for this improvement is unknown and likely multifactorial, presurgical distress, acute medical issues (eg, infection), and medications (eg, opioid analgesics) may have contributed to lower preamputation scores. The t tests using just 6 week and 4 month data reinforce the idea that although there may be variation between pre- and postsurgery, in nearly all cognitive domains there is little postsurgical variation between 6 weeks and 4 months. Appreciating that there may be transient cognitive problems prior to surgery has important implications for informed consent and other decision-making processes. From a predictive perspective, assessing cognitive performance immediately preamputation should be done with caution, taking into account potential transient and confounding factors (ie, stress pain, narcotic analgesic use) and the probability of improvement postsurgically. Learning and memory appear to be particularly vulnerable to the influence of perisurgical factors, whereas executive function, attention, and working memory abilities were comparatively less impacted. There were few associations between aggregate mean cognitive scores and health or demographic variables. At 4 months postamputation, higher levels of perceived health were associated with higher aggregate cognitive scores, consistent with other literature.⁴⁵

Better cognitive functioning has been consistently associated with superior rehabilitation outcomes after amputation,⁴⁶ including mobility,¹⁰ activities of daily living,¹⁵ and social engagement^47,48; however, the strength of the association appears to vary by the timing of the cognitive assessment and the type of outcome measure (R.M. Williams et al., unpublished data, 2013). The present study suggests that cognition has stabilized by 6 weeks postsurgery for most of the sample; this assessment point is further out from surgery than the assessment time points used in several other studies^10,15 but still precedes many important amputation and rehabilitation-related decisions (eg, prosthesis fitting, training needs).

Study limitations

In order to accommodate cognitive testing in this population, it was necessary to do a portion of assessments by phone.^49-52 Those tested via telephone at 6 weeks and 4 months reported higher levels of education than those tested in-person (13.9y vs 12.9y, P<.05); however, there were no significant differences in neuropsychological test scores between the 2 groups at any of the time points. There were no other significant differences between those assessed in person versus via telephone on other health or demographic factors. Additionally, though practice effects and regression to the mean may have impacted findings, this is rendered unlikely by the use of different test versions and the fact that scores did not improve between 6 weeks and 4 months.

The difficulty of obtaining presurgical data resulted in small sample sizes for analyses examining cognitive performance pre- to postamputation. Several factors made presurgical enrollment difficult: the window between the decision to amputate and the surgery may have been too small, potential participants were too sick or fatigued to consent prior to amputation, or potential participants were not identified until after amputation. Reassuringly, there were no significant demographic- or amputation-related differences between those who were enrolled presurgically and those enrolled postsurgically and no difference in neuropsychological test scores at 6 weeks and 4 months. This suggests that the performance of those enrolled presurgically is likely representative of a broader sample, and findings can be generalized. Significant differences over time emerged for several cognitive tasks despite limited power, lending credence to these findings. Additionally, the same patterns of improvement were seen for all participants.

As previously mentioned, participants passed a cognitive screen to be included in the study. Twenty-two percent of the initial 239 individuals were excluded from participating in the study based on their SPMSQ score. This number may be artificially low because other exclusion criteria were asked first and if excluded on the basis of earlier data (ie, this would not be their primary amputation), the SPMSQ was not administered. Therefore, participants in this study demonstrated better cognitive functioning than roughly a fifth of their peers; as such, caution should be exercised when generalizing these results to all undergoing lower limb amputation. Although one of the diagnostic criteria for dementia is impairment in ≥2 cognitive domains,⁵³ we did not look at cognition in this dichotomized fashion, though it may have facilitated comparison with other studies. This is in part because we had a cognitive exclusion criterion and also because we did not have other qualitative information to support a diagnosis of dementia.

Though our sample is typical of individuals with recent amputations seen in a rehabilitation setting, future studies are needed to examine a more diverse population of individuals with limb loss, including those with more apparent cognitive impairment at the time of baseline and those undergoing revision surgeries and/or contralateral amputations. Future studies would also be strengthened by including assessment of factors that could be expected to influence cognition, such as medication use, stress, anxiety, infection, and better assessment of pain. A control group of those undergoing a nonamputation surgery may also strengthen findings.

Conclusions

Our overarching goal was to prospectively describe cognition in a sample of individuals undergoing their first lower extremity amputation because of dysvascular disease. Cognitive performance was generally poorest immediately prior to amputation, improved by 6 weeks postamputation, and remained stable at 4 months postamputation. Change was more likely in some domains of function, suggesting that certain cognitive domains are more vulnerable to perisurgical challenges. This study addresses several gaps in the literature by including a presurgical assessment and multiple assessment points and standardized tests.

List of abbreviations:

DM

diabetes mellitus

MDE

major depressive episode

PAD

peripheral arterial disease

POCD

postoperative cognitive dysfunction

SPMSQ

Short Portable Mental Status Questionnaire